maged yahya cocomo paper
TRANSCRIPT
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 19
Impact of CMMI-Based Process Maturity Levels on
Effort Productivity and Diseconomy of Scale
Majed Alyahya Rodina Ahmad and Sai LeeDepartment of Software Engineering University of Malaya Malaysia
Abstract The Software Capability Maturity Model Integration (CMMI) has become a popular Software Process Improvement
(SPI) model for enhancing software development processes with the goal of developing high-quality software within budget
and schedule Since Software development effort can be greatly affected by the organizational process maturity level this study
examines the impact of different CMMI-based process maturity levels on effort productivity development team and
diseconomy of scale for a standard project sizes The COnstructive COst MOdel (COCOMO) is employed to compute the software development effort The percentage of change (increase or decrease) in software development effort productivity and
diseconomy of scale is employed as a measure of effectiveness for this study The results of this work demonstrate that each
higher CMMI maturity level has a considerable impact in decreasing the development effort increasing the productivity rate
and reducing the diseconomy of scale The results also indicate that the impact of CMMI-based maturity levels significantly
increases with project sizes
Keywords CMMI process maturity COCOMO II effort multipliers scale factors diseconomy of scale and productivity
rate
Received February 19 2010 accepted August 10 2010
1
Introduction
Developing software to meet functional requirementwith acceptable quality on planned schedule and
within budget is a target pursued by every softwaredevelopment organization [26] There is a widespread belief that a good software product is a result of matureand repeatable software processes which have led tomore focus on Software Process Improvement (SPI) toassist software development organizations realize its potential benefits Thus the search for reliablemethodologies ideas and innovations to enhancesoftware development continues to be an essentialfocus for both academic and industrial research Effortspent in this area has resulted in several SPI modelsand standards such as ISO 9001 [30] Six Sigma [33]and the Carnegie Mellon Software EngineeringInstitutersquos Capability Maturity Model for Software(SW-CMM) [31] as well as its most recent version theCapability Maturity Model Integration (CMMI) [7]This paper focuses on CMMI The motivation forselecting CMMI as the base of this study is that it isinfluential long-standing and often-studied standardto SPI [35] Moreover CMMI-based SPI has led toquantifiable enhancement in how processes of softwareengineering are performed [5] A cording to Jones andSoule [24] among the software process improvement
frameworks CMMI became a standard model withhigh rate of acceptanceIn this study we investigate the impact of CMMI- based process maturity levels on software developmenteffort productivity rate and diseconomy of scale for a
set of standard project sizes COCOMO II model isemployed to estimate the required effort in each levelof maturity It has a scale factor input Process
Maturity (PMAT) which is used to assess theorganizational process maturitySince the investigation presented in this work is
primarily based on CMMI and the existing COCOMOIIrsquos PMAT values were basically derived for CMMtherefore based on the results of our recent research anew set of COCOMO IIrsquos PMAT ratings values underCMMI has been derived and evaluated in [1] Thusthis study relies on the new CMMI-based PMAT ratingvalues
The rest of this research is organized as followssection 2 presents a brief background about the CMMI
maturity levels in addition to the definition of theCOCOMO Model Research motivation and hypothesisare introduced in section 3 Section 4 surveys some previous studies which are related to this workSections 5 and 6 describe the method of this work andthe evaluation measure respectively The results andrelated discussions are presented in section 7 Finallysection 8 offers some conclusions of this work and presents recommended future works
2 Background
21 CMMI Process Maturity Levels
The Software CMMI is used to rate an organizationrsquos process maturity CMMI provides a number ofrequirements that all organizations can use in setting
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 29
up the software processes used to control software product development CMMI specifies ldquowhatrdquo should be in the software process rather than ldquowhenrdquo or ldquoforhow longrdquo There are five levels of process maturitylevel 1 (lowest) to level 5 (highest) where each levelsignifies the level of performance that can be expected
from an organization For example maturity level 1organizations have ad hoc processes whereas maturitylevel 2 organizations have a basic project managementsystem in place and so on The five CMMI maturitylevels are initial (maturity level 1) managed (maturitylevel 2) defined maturity level 3) quantitativelymanaged (maturity level 4) and optimizing (maturitylevel 5) [7]
22 COCOMO II Model
The Constructive Cost Model (COCOMO) was
originally published in 1981 (COCOMO 81) [4] and became one of most popular parametric cost estimationmodels of the 1980s But in the 90s COCOMO 81faced a lot of difficulties and complications inestimating the costs of software that were developed toa new life cycle processes such as non-sequential andrapid development process models reuse-drivenapproaches and object-oriented approaches [2] ThusCOCOMO II was published initially in the Annals ofSoftware Engineering in 1995 with three sub modelsan application-composition model an early designmodel and a post-architecture model [2] COCOMO II
has as an input a set of seventeen Effort Multipliers(EM) or cost drivers which are used to adjust thenominal effort (PM) to reflect the software product being developed
221 Effort Estimation
In COCOMO II effort is expressed as Person-Months(PM) Boehm in [3] defines the person-month as ldquotheamount of time one person spends working on thesoftware development project for one monthrdquo TheCOCOMO II effort estimation equation is shown in 1
PM nominal
Where
bull A is a baseline Multiplicative Constant = 294 It isderived by the COCOMO team by calibrating to theactual effort values for the 161 projects currently inCOCOMO II database
bull The exponential factor E is discussed in the nextsection
bull The EM are used to adjust the effortbull
N is the number of effort multipliers or cost drivers
222 Scale Factors (SF)
In addition the 17 effort multipliers that are used as aninput to the COCOMO II model there is a set of fivescale factors that account for the economies anddiseconomies of scale in software development projects When there are economies of scale doubling
the software size will result in effort being less thandouble the original Whereas when diseconomies ofscale are present for a software project doubling the project size will result in more than double of theoriginal project effort being needed to complete the project [3] COCOMO II uses equation 2 to calculate ifa project has economies or diseconomies of scale
001
where B is a constant = 091 It is derived by the
COCOMO team by calibrating to the actual effortvalues for the 161 projects currently in COCOMO IIdatabase
The exponent E in (2) is an aggregation of fiveScale Factors (SF) All scale factors have rating levelsThese rating levels are Very Low (VL) Low (L) Nominal (N) High (H) Very High (VH) and ExtraHigh (XH) Each rating level has a weight W which isa quantitative value used in the COCOMO II modelAs shown in Table 1 the five COCOMO II scalefactors are Precedentedness Development FlexibilityRisk Resolution Team Cohesion and Process Maturity(PMAT) [3] The procedure for determining PMAT-the factor of interest in this study - is organized aroundthe Software Engineering Institutersquos CapabilityMaturity Model (SEI-CMM)
Table 1 Rating levels and values for COCOMO II scale factors [3]
Scale Factor
C M M L e v e l 1
( L o w e r )
C M M L e v e l 1
( U p p e r )
C M M L e v e l 2
C M M L e v e l 3
C M M L e v e l 4
C M M L e v e l 5
VL L N H VH EH
Precedentedness
(PREC)62 496 372 248 124 000
Development
Flexibility (FLEX)507 405 304 203 101 000
Risk Resolution
(RESL)707 565 424 283 141 000
Team Cohesion
(TEAM)548 438 329 219 110 000
Process Maturity
(PMAT)780 624 468 312 156 000
As can be seen in the first row of Table 1
COCOMO II uses six ratings levels of maturity Theonly difference between CMM and COCOMO IIratings levels is in the first maturity level which has been divided in COCOMO II into two halves lower
(1)
(2)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 39
half and upper half According to [8] the CMM level 1(lower half) is for organizations that rely on ldquoheroesrdquoto do the job They do not focus on processes ordocumenting lessons learned The CMM level 1 (upperhalf) is for organizations that have implemented mostof the requirements that would satisfy CMM level 2 In
CMMrsquos published definition level 1 (lower half) and(upper half) are grouped into level 1
3 Research Motivation and Hypothesis
31 Research Motivation
An increasing number of software developmentorganizations around the world have adopted CMMI toimprove their software development processes [37] Inthe literature there are numerous studies onsignificance and benefits of increasing the CMM-basedorganizational maturity levels However it is observedthat there are very limited studies on the impact and benefits of the CMMI-based process maturity It seemsCMM is still receiving much attention than CMMIeven though there are continual and widespreaddemands on the evidence about the impact and benefitsof CMMI-based process maturity from organizationsthat are adopting it [10]
Furthermore most of the available studies andresearch which focused on CMMI are case studies based on quantitative data To the best of ourknowledge qualitative studies on the effect of
increasing the CMMI maturity levels are lacked in theliterature It is believed that CMMI with its highacceptance rate as an SPI framework requires a specialinvestigation on the impact and benefits of increasingits maturity levels
32 Research Hypothesis
The hypothesis of the work presented here is thatincreasing the level of CMMI-based process maturitywill results in the following
1 A decrease in the software development effort
2
An increase in the productivity rate3 A reduce the in diseconomy of scale
As stated earlier this study relies on the new CMMI- based PMAT rating values which are shown in Table2
Table 2 The new PMAT rating values
PMAT
Description
CMMI
Level 1
(Lower)
CMMI
Level 1
(Upper)
CMMI
Level 2
CMMI
Level 3
CMMI
Level 4
CMMI
Level 5
Rating
Levels
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
New PMATValues
755 571 381 208 103 000
Since COCOMO II considers CMM level 2 as thenominal rating for PMAT the same consideration is
for CMMI level 2 Thus the percentage of increase ordecrease in effort productivity and diseconomy ofscale will be compared against the nominal CMMI- based PMAT ratings
4
Related Literature
Numerous researches and case studies have shownmany benefits of enhancing organizational processmaturity by using different assessment approaches [622 23 27 36] Girish et al [18] conducted anempirical study to investigate the effects of CMM ontwo critical factors in Information Systems (IS)implementation strategy which are project performance and software quality They claimed thatCMM levels are associated with IS implementationstrategies and higher CMM levels are relate to higher project performance and software quality that lead to
noticeable reduction in software development effortand schedule From a review of seventeen publishedarticles Galin and Avrahami [15] explored CMM- based benefits such as defects rework schedule productivity error defection effectiveness and ReturnOn Investment (ROI) concluding that a goodinvestment in CMM programs leads to enhancedsoftware development and maintenance Diaz and King[11] claimed that increase in CMM process maturityresults in an improvement in quality phasecontainment productivity and rework In order toexplore the impact of process maturity on software
development effort and based on CMM with the aid of161-project sample Clark in [8] isolated the effects oneffort of process maturity versus effects of otherfactors and found that an increasing of oneorganizational process maturity level can result in areduction in software development effort by 4 to11 But this reduction seemed like a generalizationacross all five levels of CMM process maturity iethe percentage of effort reduction is not the sameamong all levels Memon has reported in [29] thatCMM maturity levels considerably influences thesoftware development effort and productivity El
Emam and Goldenson [13] in an comprehensivereview of studies and publications on theimplementation of SPI methodologies includingCMM reported qualitative performance improvementsin terms of higher quality higher productivityimproved ability to meet development schedulesDonald et al [12] conducted an empirical research tofind out the relationship between quality of the products organizational process maturity developmenteffort and project schedule Their findings indicatedthat process maturity has an effect in reducing softwaredevelopment schedule and effort Another survey- based study of individuals from SW-CMM-assessedsoftware organizations revealed that higher maturityorganizations are associated with better performanceincluding the ability to meet budget and schedule as
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 49
well as increase staff productivity product quality andcustomer satisfaction [20] Herbsleb and Goldenson in[21] showed solid evidence in a sample of 61 softwareorganizations that high CMM-based software processmaturity is associated with high performance
Despite the numerous studies that have investigated
the performance assessment results of CMM-basedsoftware process maturity and its impact on softwaredevelopment effort there are still very limited workson the overall benefits of CMMI-based software process maturity [37] Case studies have also shown benefits from CMMI-based software process maturityin [14 16 17 19 25 28 32 34 37] Goldenson andGibson in [19] reported some great and crediblequantitative evidence that CMMI-based software process maturity can help an organization achievehigher quality products and better project performancewith lower cost and decreased project effort Due to the
limitation of the performance results provided in [19]Gibson et al [17] continued the assessment performance of CMMI-based software processimprovement and provides empirical tangible evidenceabout the performance results that can achieve as anoutcome of CMMI-based process improvement Theyreported ldquoThere now is ample evidence that processimprovement using the CMMI Product Suite can resultin marked improvements in schedule and cost performance product quality return on investment andother measures of performance outcomerdquo
5
Research Method
In order to investigate the impact of CMMI-based process maturity level on a variety of software productsizes it is more recommended to classify the softwaresizes in an appropriate manner since size is consideredthe most influential factor in predicting effort of thesoftware product [9] Boehm in [4] has classified thesoftware product sizes as small intermediate mediumlarge and very large See Table 3
Table 3 Software size classification according to [4]
Classification Size (KLOC)
Small (S) 2
Intermediate (I) 8
Medium (M) 32
Large (L) 128
Very Large (VL) 512
The scale factor PMAT is used here to capture theimpact of different process maturity levels on softwaredevelopment effort for standard size projects classifiedabove
51 Effort Estimation
The basic idea in our research method is quantifyingthe impact of PMAT versus other factors that influence
the software development effort To do so we separatethe impact of PMAT from other factors because whendifferent kinds of improvements are carried outconcurrently in the organization project managers willhave no idea on how to determine the amount ofimprovement gained from process maturity with the
presence of other factors [8] In this contextCOCOMO II model is employed in order to estimatethe effort of software development Subsequently alleffort multipliers are set to be nominal ie the valueof 1 Furthermore as described in Table 4 all scalefactors except PMAT (the one related to the processmaturity) are also assumed to be nominal The effortmultipliers and scale factors (except PMAT) are set totheir nominal values in order to isolate their potentialeffects on software development effort As an examplefor a nominal PMAT rating and a standard large size project by substituting values in equations 1 and 2 we
get 294128 58521
Table 4 Nominal values for all scale factors in all rating levels(except CMMI-based PAMT)
Scale
Factor
(SF)
CMMI
Level 1
(lower)
CMMI
Level 1
(upper)
CMMI
Level 2
CMMI
Level 3
CMMI
Level 4
CMM I
Level 5
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
Precedentedness
(PREC)372 372 372 372 372 372
DevelopmentFlexibility (FLEX) 304 304 304 304 304 304
Risk
Resolution (RESL)424 424 424 424 424 424
Team
Cohesion
(TEAM)
329 329 329 329 329 329
New
Process
Maturity
(PMAT)
755 571 381 208 103 000
Summation
of All SF2184 2000 1810 1637 1532 1429
52 Productivity RateIn order to test our hypothesis which is increasing thelevel of CMMI-based process maturity increases the productivity rate equation 3 is applied for eachestimated effort
where SIZE is the standard size used and it ismeasured in this formula by thousand lines of codes(KLOC) and EFFORT is the effort estimated in eachPMAT level for all standard sizes
As an example for the productivity for nominalPMAT rating and standard large size project equation3 will be applied to the effort produced in the previoussection
(3)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 59
12858521 21872
53 Diseconomy of Scale
Diseconomy of scale refers to relatively more increasein effort as compared to the increase in size of a
software product That is doubling the project size willresult in more than double of the original project effort being needed to complete the project
To make this concept more clear Memon [29]suggested dividing the standard project sizes discussedearlier by the small size (2 KLOC) and then calledfrom small to intermediate (from S to I) from small tomedium (from S to M) from small to large (from S toL) and from small to very large (from S to VL)Similarly their corresponding efforts are also divided by the effort of small size in order to visualize theeffect of diseconomy of scale see Table 5
Table 5 Software sizes ratio
Classification Size Ration
From S to I 4From S to M 16From S to L 64
From S to VL 256
6 Evaluation Measure
In order to evaluate and compare our proposed resultsmeasures of effectiveness are required In this study
the percentage change in software development effort productivity rate and diseconomy of scale are used asthe primary Measures Of Effectiveness (MOE) The percent change means either increase or decrease ineffort productivity rate and diseconomy of scale Thiscriterion measure will determine the magnitude of theeffect of different process maturity levels ondevelopment effort for all standard sizes of software projects To compute these percentages we willassume the nominal rating of PMAT as the base caseHence the percent of change in effort productivityand diseconomy of scale is measured by using equation4
where Parameter refers to the computed value ofeffort productivity or diseconomy of scale at a particular PMAT rating Whereas Parameter nominal referto the nominal value of the same parameter
A combination of positive and negative changes will be seen in the resulted values A negative valueindicates percentage reduction in the parameter valueand a positive one show percentage increase in the
parameter value
7 Results and Discussions
71 Effort
After applying our method COCOMO II has estimatedthe effort for all standard project sizes in each PAMTrating Table 6 shows the resulted effort Table 7
shows the percentage change in effort in each processmaturity level for all standard size projects and Figure1 is a visual representation of Table 7
Table 6 Estimated effort in all CMMI-based PMAT ratings for allstandard sizes
Project Effort (PM) Based on PMAT Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 643 635 626 619 614 610
Intermediate 8 3072 2956 2842 2742 2682 2625
Medium 32 14681 13774 12896 12146 11712 11301
Large 128 70165 64173 58521 53809 51137 48644
Very Large 512 335342 298976 265559 238391 223277 209381
Table 7 Percent change of effort in all CMMI-based PMAT ratingsfor all standard sizes
Project Average Change in Effort
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 263 133 000 -119 -191 -261
Intermediate 8 809 403 000 -353 -562 -762
Medium 32 1384 681 000 -582 -919 -1237
Large 128 1990 966 000 -805 -1262 -1688
Very Large 512 2628 1258 000 -1023 -1592 -2115
Figure 1 Percent change of effort in all PMAT ratings for allstandard sizes
As seen in Table 6 the results indicate that for eachimprovement in the PMAT rating there is a decreasein the required software development effort It can benoticed that these improvements in the effort are moreconsiderable for larger projects size than for smallersizes Also the percentage change shown in Table 7
and graphically in Figure 1 obviously gives the sameindications The percent change in required effortvaries from increase of (263) for very low rating ofPMAT to a decrease of 261 for a rating of extra high(a total change of 524) for small size projects
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983109 983142 983142 983151 983154 983156
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983109983142983142983151983154983156 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141983154 983149983141 983140983145983137983156983141 983117983141983140983145983157983149 983116 983137983154983143983141 983126983141983154 983161 983116983137983154 983143983141
(4)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 69
whereas for very large projects the percent change inrequired effort varies from an increase of (2628) forvery low rating of PMAT to a decrease of 2115 for arating of extra high (a total change of 4743)
72 Productivity Rate
Table 8 and Figure 2 show the resulted productivityWhile the percent change in productivity in each process maturity level for all standard size projects areshown in Table 9 and Figure 3
Table 8 Productivity rate in all CMMI-based PMAT ratings for allstandard sizes
ProjectProductivity (KLOCPM) Based on ISF-PMAT
Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 311 315 31934 323 326 328
Intermediate 8 260 271 28150 292 298 305Medium 32 218 232 24813 263 273 283
Large 128 182 199 21872 238 250 263
Very Large 512 153 171 19280 215 229 245
Figure 2 Productivity rate in all PMAT ratings for all standardsizes
As can be seen in Table 8 and visually in Figure 2there is a considerable growth in the productivity ratein all PMAT levels They also give a clear indicationthat the productivity rate changes (increases) more
rapidly for larger projects as compared to smaller projects
Table 9 Percent change of productivity in all CMMI-based PMATratings for all standard sizes
Project Average Change in Productivity
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 -256 -131 000 121 195 268
Intermediate 8 -748 -387 000 366 595 824
Medium 32 -1216 -637 000 618 1011 1412
Large 128 -1660 -881 000 876 1444 2031
Very Large 512 -2081 -1118 000 1140 1894 2683
As shown in Table 9 and graphically in Figure 3 the percent change in productivity rate varies fromdecrease of (256) for very low rating of PMAT to an
increase of 268 for a rating of extra high (a totalchange of 524) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from a decrease of (2081) for very lowrating of PMAT to an increase of 2683 for a ratingof extra high (a total change of 4764)
Figure 3 Percent change of productivity in all PMAT ratings for allstandard sizes
73 Diseconomy of Scale
When the standard sizes were divided by the small size(2 KLOC) similarly their corresponding efforts arealso divided by the effort of small size to visualize theeffect of diseconomy of scale as shown in Table 10 andFigure 4 To visualize the effect of diseconomy ofscale the size ratio has been plotted in Figure 4 The percent change in diseconomy of scale in each process
maturity level for all standard size projects are shownin Table 11 and Figure 5
Table 10 Diseconomy of scale in all CMMI-based PMAT ratingsfor all standard sizes
ProjectDiseconomy of Scale Based on CMMI-Based
PMAT Ratings and Project Size
ClassificationSizeRatio
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 478 466 454 443 437 430
From S to M 16 2284 2171 2059 1963 1906 1853
From S to L 64 10917 10113 9344 8696 8324 7975
From S to VL 256 52174 47114 42402 38524 36345 34327
Figure 4 Diseconomy of scale in all PMAT ratings for all standardsizes
Table 10 and Figure 4 show that diseconomy ofscale improves desirably with improvement in PMAT
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983154983151983140983157983139983156983145983158983145983156983161 983122983137983156983141983155 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141 983154983149983141983140983145983137983156983141 983117983141 983140983145983157983149 983116 983137983154 983143983141 983126983141 983154983161 983116983137983154983143983141
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150 983137983148 983112983145 983143983144 983126983141 983154983161 983112983145 983143983144 983109983160 983156983154 983137 983112983145 983143983144
983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983120983154983151983140983157983139983156983145983158983145983156983161 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983141 983113983150983156983141983154983149983141 983140983145983137983156983141 983117983141 983140983145983157983149 983116983137983154983143983141 983126983141 983154983161 983116983137983154983143983141
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983093983088983088
983094983088983088
983123983145983162983141
983122983137983156983145983151
983126983141983154983161
983116983151983159
983116 983151983159 983118983151 983149983145983150 983137983148 983112983145983143983144 983126983141 983154983161
983112983145983143983144
983109983160983156983154983137
983112983145983143983144
983122 983137 983156 983145 983151 983151 983142 983109 983142 983142 983151 983154 983156 983087 983123 983145 983162 983141
983120983117983105983124 983148983141983158983141983148 983154983137983156983145983150983143 983087 983123983145983162983141 983080983115983116983119983107983081
983108983145983155983141983139983151983150983151983149983161 983151983142 983123983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 79
rating levels They also give an obvious andconsiderable indication that the productivity ratechanges (increases) more rapidly for larger projects ascompared to smaller projects
Table 11 Percent Change of Diseconomy of Scale in all CMMI- based PMAT Ratings for all Standard sizes
Project Average Change in Diseconomy of Scale
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 532 267 000 -237 -378 -514
From S to M 16 1093 541 000 -468 -742 -1002
From S to L 64 1683 822 000 -694 -1092 -1465
From S to VL 256 2305 1111 000 -915 -1429 -1904
Figure 5 Percent change of diseconomy of scale in all PMATratings for all standard sizes
As shown in Table 11 and graphically in Figure 5
the percent change in diseconomy of scale varies froman increase of (532) for very low rating of PMAT toa decrease of 514 for a rating of extra high (a totalchange of 1046) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from an increase of (2305) for very lowrating of PMAT to a decrease of 1904 for a rating ofextra high (a total change of 4209)
8 Conclusions and Future Work
This paper investigates the impact of CMMI-based
process maturity levels on software developmenteffort productivity rate and diseconomy of scale for aset of standard project sizes The result of thisinvestigation has boosted our hypotheses and revealedthat the process maturity rating levels based on CMMIconsiderably affect the software development effort interms of productivity and diseconomy of scale ie foreach higher level of PMAT there is a decrease ineffort required increase in productivity rate andreduction in diseconomy of scale although the percentage of (increasedecrease) is not uniform amonglevels The results also demonstrate that this effect is
more considerable and significant for larger projects ascompared to smaller projects This necessitates theadoption of CMMI-based process improvement forsoftware organizations that are developing projects of
large sizes As a result of investigating the impact of process maturity on effort it seems reasonable tosupport the suggestion that PMAT is a significant inputto software cost estimation models
Future work in the area of CMMI-based processmaturity requires collecting historical data for each of
the 22 process areas used in CMMI in order to examinewhich process area has most impact on effort productivity and diseconomy of scale In additionunlike SW-CMM CMMI has two differentrepresentations Staged and Continuous This studyfocused on the staged representation therefore furtherresearch is recommended to be conducted on theimpact of CMMI-based Process capability from thecontinuous representation perspective
References
[1]
Alyahya A Ahmad R and Lee S ldquoImpact ofCMMI Based Software Process Maturity onCOCOMO IIs Effort Estimationrdquo International Arab Journal of Information Technology vol 7no 2 pp 129-137 2010
[2] Boehm B Clark B Horowitz E Westland CMadachy R and Selby R ldquoCost Models forFuture Software Life Cycle ProcessesCOCOMO 20rdquo in Proceedings of SpecialVolume on Software Process and Product Measurement Amsterdam pp 45-60 1995
[3] Boehm B Horowitz E Madachy R Reifer D
Clark B Steece B Brown A Chulani S andAbts C Software Cost Estimation withCOCOMO II Prentice Hall 2000
[4] Boehm B Software Engineering EconomicsPrentice Hall 1981
[5] Bollinger T and McGowan C ldquoA Critical Lookat Software Capability Evaluationsrdquo IEEESoftware vol 26 no 5 pp 80-83 2009
[6] Butler K ldquoThe Economic Benefits of SoftwareProcess Improvementrdquo in Proceedings of Crosstalk Hill AFB Ogden pp 14-17 1995
[7] Chrissis M Konrad M and Shrum S CMMI
Guidelines for Process Integration and Product Improvement 3ed Addison-Wesley 2003
[8] Clark B ldquoQuantifying the Effects of ProcessImprovement on Effortrdquo IEEE Softwarevol 17 no 6 pp 65-70 2000
[9] Clark B ldquoThe Effects of Software ProcessMaturity on Software Development Effortrdquo PhD Dissertation University of Southern California1997
[10] Damian D Zowghi D Vaidyanathasamy L andPal Y ldquoAn Industrial Experience in ProcessImprovement An Early Assessment at theAustralian Center for Unisys Softwarerdquo in Proceedings of International Symposium on Empirical Software Engineering (ISESE) pp111-123 2002
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144
983108 983145 983155 983141 983139 983151 983150 983151 983149 983161 983151 983140 983123 983139 983137 983148 983141
983120983154983151983139983141983155983155 983117983137983156983157983154983145983156983161 983116 983141983158983141983148
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983108983145983155983141983139983151983150983151983149983161 983151983140 983155983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 29
up the software processes used to control software product development CMMI specifies ldquowhatrdquo should be in the software process rather than ldquowhenrdquo or ldquoforhow longrdquo There are five levels of process maturitylevel 1 (lowest) to level 5 (highest) where each levelsignifies the level of performance that can be expected
from an organization For example maturity level 1organizations have ad hoc processes whereas maturitylevel 2 organizations have a basic project managementsystem in place and so on The five CMMI maturitylevels are initial (maturity level 1) managed (maturitylevel 2) defined maturity level 3) quantitativelymanaged (maturity level 4) and optimizing (maturitylevel 5) [7]
22 COCOMO II Model
The Constructive Cost Model (COCOMO) was
originally published in 1981 (COCOMO 81) [4] and became one of most popular parametric cost estimationmodels of the 1980s But in the 90s COCOMO 81faced a lot of difficulties and complications inestimating the costs of software that were developed toa new life cycle processes such as non-sequential andrapid development process models reuse-drivenapproaches and object-oriented approaches [2] ThusCOCOMO II was published initially in the Annals ofSoftware Engineering in 1995 with three sub modelsan application-composition model an early designmodel and a post-architecture model [2] COCOMO II
has as an input a set of seventeen Effort Multipliers(EM) or cost drivers which are used to adjust thenominal effort (PM) to reflect the software product being developed
221 Effort Estimation
In COCOMO II effort is expressed as Person-Months(PM) Boehm in [3] defines the person-month as ldquotheamount of time one person spends working on thesoftware development project for one monthrdquo TheCOCOMO II effort estimation equation is shown in 1
PM nominal
Where
bull A is a baseline Multiplicative Constant = 294 It isderived by the COCOMO team by calibrating to theactual effort values for the 161 projects currently inCOCOMO II database
bull The exponential factor E is discussed in the nextsection
bull The EM are used to adjust the effortbull
N is the number of effort multipliers or cost drivers
222 Scale Factors (SF)
In addition the 17 effort multipliers that are used as aninput to the COCOMO II model there is a set of fivescale factors that account for the economies anddiseconomies of scale in software development projects When there are economies of scale doubling
the software size will result in effort being less thandouble the original Whereas when diseconomies ofscale are present for a software project doubling the project size will result in more than double of theoriginal project effort being needed to complete the project [3] COCOMO II uses equation 2 to calculate ifa project has economies or diseconomies of scale
001
where B is a constant = 091 It is derived by the
COCOMO team by calibrating to the actual effortvalues for the 161 projects currently in COCOMO IIdatabase
The exponent E in (2) is an aggregation of fiveScale Factors (SF) All scale factors have rating levelsThese rating levels are Very Low (VL) Low (L) Nominal (N) High (H) Very High (VH) and ExtraHigh (XH) Each rating level has a weight W which isa quantitative value used in the COCOMO II modelAs shown in Table 1 the five COCOMO II scalefactors are Precedentedness Development FlexibilityRisk Resolution Team Cohesion and Process Maturity(PMAT) [3] The procedure for determining PMAT-the factor of interest in this study - is organized aroundthe Software Engineering Institutersquos CapabilityMaturity Model (SEI-CMM)
Table 1 Rating levels and values for COCOMO II scale factors [3]
Scale Factor
C M M L e v e l 1
( L o w e r )
C M M L e v e l 1
( U p p e r )
C M M L e v e l 2
C M M L e v e l 3
C M M L e v e l 4
C M M L e v e l 5
VL L N H VH EH
Precedentedness
(PREC)62 496 372 248 124 000
Development
Flexibility (FLEX)507 405 304 203 101 000
Risk Resolution
(RESL)707 565 424 283 141 000
Team Cohesion
(TEAM)548 438 329 219 110 000
Process Maturity
(PMAT)780 624 468 312 156 000
As can be seen in the first row of Table 1
COCOMO II uses six ratings levels of maturity Theonly difference between CMM and COCOMO IIratings levels is in the first maturity level which has been divided in COCOMO II into two halves lower
(1)
(2)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 39
half and upper half According to [8] the CMM level 1(lower half) is for organizations that rely on ldquoheroesrdquoto do the job They do not focus on processes ordocumenting lessons learned The CMM level 1 (upperhalf) is for organizations that have implemented mostof the requirements that would satisfy CMM level 2 In
CMMrsquos published definition level 1 (lower half) and(upper half) are grouped into level 1
3 Research Motivation and Hypothesis
31 Research Motivation
An increasing number of software developmentorganizations around the world have adopted CMMI toimprove their software development processes [37] Inthe literature there are numerous studies onsignificance and benefits of increasing the CMM-basedorganizational maturity levels However it is observedthat there are very limited studies on the impact and benefits of the CMMI-based process maturity It seemsCMM is still receiving much attention than CMMIeven though there are continual and widespreaddemands on the evidence about the impact and benefitsof CMMI-based process maturity from organizationsthat are adopting it [10]
Furthermore most of the available studies andresearch which focused on CMMI are case studies based on quantitative data To the best of ourknowledge qualitative studies on the effect of
increasing the CMMI maturity levels are lacked in theliterature It is believed that CMMI with its highacceptance rate as an SPI framework requires a specialinvestigation on the impact and benefits of increasingits maturity levels
32 Research Hypothesis
The hypothesis of the work presented here is thatincreasing the level of CMMI-based process maturitywill results in the following
1 A decrease in the software development effort
2
An increase in the productivity rate3 A reduce the in diseconomy of scale
As stated earlier this study relies on the new CMMI- based PMAT rating values which are shown in Table2
Table 2 The new PMAT rating values
PMAT
Description
CMMI
Level 1
(Lower)
CMMI
Level 1
(Upper)
CMMI
Level 2
CMMI
Level 3
CMMI
Level 4
CMMI
Level 5
Rating
Levels
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
New PMATValues
755 571 381 208 103 000
Since COCOMO II considers CMM level 2 as thenominal rating for PMAT the same consideration is
for CMMI level 2 Thus the percentage of increase ordecrease in effort productivity and diseconomy ofscale will be compared against the nominal CMMI- based PMAT ratings
4
Related Literature
Numerous researches and case studies have shownmany benefits of enhancing organizational processmaturity by using different assessment approaches [622 23 27 36] Girish et al [18] conducted anempirical study to investigate the effects of CMM ontwo critical factors in Information Systems (IS)implementation strategy which are project performance and software quality They claimed thatCMM levels are associated with IS implementationstrategies and higher CMM levels are relate to higher project performance and software quality that lead to
noticeable reduction in software development effortand schedule From a review of seventeen publishedarticles Galin and Avrahami [15] explored CMM- based benefits such as defects rework schedule productivity error defection effectiveness and ReturnOn Investment (ROI) concluding that a goodinvestment in CMM programs leads to enhancedsoftware development and maintenance Diaz and King[11] claimed that increase in CMM process maturityresults in an improvement in quality phasecontainment productivity and rework In order toexplore the impact of process maturity on software
development effort and based on CMM with the aid of161-project sample Clark in [8] isolated the effects oneffort of process maturity versus effects of otherfactors and found that an increasing of oneorganizational process maturity level can result in areduction in software development effort by 4 to11 But this reduction seemed like a generalizationacross all five levels of CMM process maturity iethe percentage of effort reduction is not the sameamong all levels Memon has reported in [29] thatCMM maturity levels considerably influences thesoftware development effort and productivity El
Emam and Goldenson [13] in an comprehensivereview of studies and publications on theimplementation of SPI methodologies includingCMM reported qualitative performance improvementsin terms of higher quality higher productivityimproved ability to meet development schedulesDonald et al [12] conducted an empirical research tofind out the relationship between quality of the products organizational process maturity developmenteffort and project schedule Their findings indicatedthat process maturity has an effect in reducing softwaredevelopment schedule and effort Another survey- based study of individuals from SW-CMM-assessedsoftware organizations revealed that higher maturityorganizations are associated with better performanceincluding the ability to meet budget and schedule as
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 49
well as increase staff productivity product quality andcustomer satisfaction [20] Herbsleb and Goldenson in[21] showed solid evidence in a sample of 61 softwareorganizations that high CMM-based software processmaturity is associated with high performance
Despite the numerous studies that have investigated
the performance assessment results of CMM-basedsoftware process maturity and its impact on softwaredevelopment effort there are still very limited workson the overall benefits of CMMI-based software process maturity [37] Case studies have also shown benefits from CMMI-based software process maturityin [14 16 17 19 25 28 32 34 37] Goldenson andGibson in [19] reported some great and crediblequantitative evidence that CMMI-based software process maturity can help an organization achievehigher quality products and better project performancewith lower cost and decreased project effort Due to the
limitation of the performance results provided in [19]Gibson et al [17] continued the assessment performance of CMMI-based software processimprovement and provides empirical tangible evidenceabout the performance results that can achieve as anoutcome of CMMI-based process improvement Theyreported ldquoThere now is ample evidence that processimprovement using the CMMI Product Suite can resultin marked improvements in schedule and cost performance product quality return on investment andother measures of performance outcomerdquo
5
Research Method
In order to investigate the impact of CMMI-based process maturity level on a variety of software productsizes it is more recommended to classify the softwaresizes in an appropriate manner since size is consideredthe most influential factor in predicting effort of thesoftware product [9] Boehm in [4] has classified thesoftware product sizes as small intermediate mediumlarge and very large See Table 3
Table 3 Software size classification according to [4]
Classification Size (KLOC)
Small (S) 2
Intermediate (I) 8
Medium (M) 32
Large (L) 128
Very Large (VL) 512
The scale factor PMAT is used here to capture theimpact of different process maturity levels on softwaredevelopment effort for standard size projects classifiedabove
51 Effort Estimation
The basic idea in our research method is quantifyingthe impact of PMAT versus other factors that influence
the software development effort To do so we separatethe impact of PMAT from other factors because whendifferent kinds of improvements are carried outconcurrently in the organization project managers willhave no idea on how to determine the amount ofimprovement gained from process maturity with the
presence of other factors [8] In this contextCOCOMO II model is employed in order to estimatethe effort of software development Subsequently alleffort multipliers are set to be nominal ie the valueof 1 Furthermore as described in Table 4 all scalefactors except PMAT (the one related to the processmaturity) are also assumed to be nominal The effortmultipliers and scale factors (except PMAT) are set totheir nominal values in order to isolate their potentialeffects on software development effort As an examplefor a nominal PMAT rating and a standard large size project by substituting values in equations 1 and 2 we
get 294128 58521
Table 4 Nominal values for all scale factors in all rating levels(except CMMI-based PAMT)
Scale
Factor
(SF)
CMMI
Level 1
(lower)
CMMI
Level 1
(upper)
CMMI
Level 2
CMMI
Level 3
CMMI
Level 4
CMM I
Level 5
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
Precedentedness
(PREC)372 372 372 372 372 372
DevelopmentFlexibility (FLEX) 304 304 304 304 304 304
Risk
Resolution (RESL)424 424 424 424 424 424
Team
Cohesion
(TEAM)
329 329 329 329 329 329
New
Process
Maturity
(PMAT)
755 571 381 208 103 000
Summation
of All SF2184 2000 1810 1637 1532 1429
52 Productivity RateIn order to test our hypothesis which is increasing thelevel of CMMI-based process maturity increases the productivity rate equation 3 is applied for eachestimated effort
where SIZE is the standard size used and it ismeasured in this formula by thousand lines of codes(KLOC) and EFFORT is the effort estimated in eachPMAT level for all standard sizes
As an example for the productivity for nominalPMAT rating and standard large size project equation3 will be applied to the effort produced in the previoussection
(3)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 59
12858521 21872
53 Diseconomy of Scale
Diseconomy of scale refers to relatively more increasein effort as compared to the increase in size of a
software product That is doubling the project size willresult in more than double of the original project effort being needed to complete the project
To make this concept more clear Memon [29]suggested dividing the standard project sizes discussedearlier by the small size (2 KLOC) and then calledfrom small to intermediate (from S to I) from small tomedium (from S to M) from small to large (from S toL) and from small to very large (from S to VL)Similarly their corresponding efforts are also divided by the effort of small size in order to visualize theeffect of diseconomy of scale see Table 5
Table 5 Software sizes ratio
Classification Size Ration
From S to I 4From S to M 16From S to L 64
From S to VL 256
6 Evaluation Measure
In order to evaluate and compare our proposed resultsmeasures of effectiveness are required In this study
the percentage change in software development effort productivity rate and diseconomy of scale are used asthe primary Measures Of Effectiveness (MOE) The percent change means either increase or decrease ineffort productivity rate and diseconomy of scale Thiscriterion measure will determine the magnitude of theeffect of different process maturity levels ondevelopment effort for all standard sizes of software projects To compute these percentages we willassume the nominal rating of PMAT as the base caseHence the percent of change in effort productivityand diseconomy of scale is measured by using equation4
where Parameter refers to the computed value ofeffort productivity or diseconomy of scale at a particular PMAT rating Whereas Parameter nominal referto the nominal value of the same parameter
A combination of positive and negative changes will be seen in the resulted values A negative valueindicates percentage reduction in the parameter valueand a positive one show percentage increase in the
parameter value
7 Results and Discussions
71 Effort
After applying our method COCOMO II has estimatedthe effort for all standard project sizes in each PAMTrating Table 6 shows the resulted effort Table 7
shows the percentage change in effort in each processmaturity level for all standard size projects and Figure1 is a visual representation of Table 7
Table 6 Estimated effort in all CMMI-based PMAT ratings for allstandard sizes
Project Effort (PM) Based on PMAT Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 643 635 626 619 614 610
Intermediate 8 3072 2956 2842 2742 2682 2625
Medium 32 14681 13774 12896 12146 11712 11301
Large 128 70165 64173 58521 53809 51137 48644
Very Large 512 335342 298976 265559 238391 223277 209381
Table 7 Percent change of effort in all CMMI-based PMAT ratingsfor all standard sizes
Project Average Change in Effort
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 263 133 000 -119 -191 -261
Intermediate 8 809 403 000 -353 -562 -762
Medium 32 1384 681 000 -582 -919 -1237
Large 128 1990 966 000 -805 -1262 -1688
Very Large 512 2628 1258 000 -1023 -1592 -2115
Figure 1 Percent change of effort in all PMAT ratings for allstandard sizes
As seen in Table 6 the results indicate that for eachimprovement in the PMAT rating there is a decreasein the required software development effort It can benoticed that these improvements in the effort are moreconsiderable for larger projects size than for smallersizes Also the percentage change shown in Table 7
and graphically in Figure 1 obviously gives the sameindications The percent change in required effortvaries from increase of (263) for very low rating ofPMAT to a decrease of 261 for a rating of extra high(a total change of 524) for small size projects
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983109 983142 983142 983151 983154 983156
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983109983142983142983151983154983156 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141983154 983149983141 983140983145983137983156983141 983117983141983140983145983157983149 983116 983137983154983143983141 983126983141983154 983161 983116983137983154 983143983141
(4)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 69
whereas for very large projects the percent change inrequired effort varies from an increase of (2628) forvery low rating of PMAT to a decrease of 2115 for arating of extra high (a total change of 4743)
72 Productivity Rate
Table 8 and Figure 2 show the resulted productivityWhile the percent change in productivity in each process maturity level for all standard size projects areshown in Table 9 and Figure 3
Table 8 Productivity rate in all CMMI-based PMAT ratings for allstandard sizes
ProjectProductivity (KLOCPM) Based on ISF-PMAT
Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 311 315 31934 323 326 328
Intermediate 8 260 271 28150 292 298 305Medium 32 218 232 24813 263 273 283
Large 128 182 199 21872 238 250 263
Very Large 512 153 171 19280 215 229 245
Figure 2 Productivity rate in all PMAT ratings for all standardsizes
As can be seen in Table 8 and visually in Figure 2there is a considerable growth in the productivity ratein all PMAT levels They also give a clear indicationthat the productivity rate changes (increases) more
rapidly for larger projects as compared to smaller projects
Table 9 Percent change of productivity in all CMMI-based PMATratings for all standard sizes
Project Average Change in Productivity
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 -256 -131 000 121 195 268
Intermediate 8 -748 -387 000 366 595 824
Medium 32 -1216 -637 000 618 1011 1412
Large 128 -1660 -881 000 876 1444 2031
Very Large 512 -2081 -1118 000 1140 1894 2683
As shown in Table 9 and graphically in Figure 3 the percent change in productivity rate varies fromdecrease of (256) for very low rating of PMAT to an
increase of 268 for a rating of extra high (a totalchange of 524) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from a decrease of (2081) for very lowrating of PMAT to an increase of 2683 for a ratingof extra high (a total change of 4764)
Figure 3 Percent change of productivity in all PMAT ratings for allstandard sizes
73 Diseconomy of Scale
When the standard sizes were divided by the small size(2 KLOC) similarly their corresponding efforts arealso divided by the effort of small size to visualize theeffect of diseconomy of scale as shown in Table 10 andFigure 4 To visualize the effect of diseconomy ofscale the size ratio has been plotted in Figure 4 The percent change in diseconomy of scale in each process
maturity level for all standard size projects are shownin Table 11 and Figure 5
Table 10 Diseconomy of scale in all CMMI-based PMAT ratingsfor all standard sizes
ProjectDiseconomy of Scale Based on CMMI-Based
PMAT Ratings and Project Size
ClassificationSizeRatio
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 478 466 454 443 437 430
From S to M 16 2284 2171 2059 1963 1906 1853
From S to L 64 10917 10113 9344 8696 8324 7975
From S to VL 256 52174 47114 42402 38524 36345 34327
Figure 4 Diseconomy of scale in all PMAT ratings for all standardsizes
Table 10 and Figure 4 show that diseconomy ofscale improves desirably with improvement in PMAT
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983154983151983140983157983139983156983145983158983145983156983161 983122983137983156983141983155 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141 983154983149983141983140983145983137983156983141 983117983141 983140983145983157983149 983116 983137983154 983143983141 983126983141 983154983161 983116983137983154983143983141
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150 983137983148 983112983145 983143983144 983126983141 983154983161 983112983145 983143983144 983109983160 983156983154 983137 983112983145 983143983144
983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983120983154983151983140983157983139983156983145983158983145983156983161 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983141 983113983150983156983141983154983149983141 983140983145983137983156983141 983117983141 983140983145983157983149 983116983137983154983143983141 983126983141 983154983161 983116983137983154983143983141
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983093983088983088
983094983088983088
983123983145983162983141
983122983137983156983145983151
983126983141983154983161
983116983151983159
983116 983151983159 983118983151 983149983145983150 983137983148 983112983145983143983144 983126983141 983154983161
983112983145983143983144
983109983160983156983154983137
983112983145983143983144
983122 983137 983156 983145 983151 983151 983142 983109 983142 983142 983151 983154 983156 983087 983123 983145 983162 983141
983120983117983105983124 983148983141983158983141983148 983154983137983156983145983150983143 983087 983123983145983162983141 983080983115983116983119983107983081
983108983145983155983141983139983151983150983151983149983161 983151983142 983123983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 79
rating levels They also give an obvious andconsiderable indication that the productivity ratechanges (increases) more rapidly for larger projects ascompared to smaller projects
Table 11 Percent Change of Diseconomy of Scale in all CMMI- based PMAT Ratings for all Standard sizes
Project Average Change in Diseconomy of Scale
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 532 267 000 -237 -378 -514
From S to M 16 1093 541 000 -468 -742 -1002
From S to L 64 1683 822 000 -694 -1092 -1465
From S to VL 256 2305 1111 000 -915 -1429 -1904
Figure 5 Percent change of diseconomy of scale in all PMATratings for all standard sizes
As shown in Table 11 and graphically in Figure 5
the percent change in diseconomy of scale varies froman increase of (532) for very low rating of PMAT toa decrease of 514 for a rating of extra high (a totalchange of 1046) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from an increase of (2305) for very lowrating of PMAT to a decrease of 1904 for a rating ofextra high (a total change of 4209)
8 Conclusions and Future Work
This paper investigates the impact of CMMI-based
process maturity levels on software developmenteffort productivity rate and diseconomy of scale for aset of standard project sizes The result of thisinvestigation has boosted our hypotheses and revealedthat the process maturity rating levels based on CMMIconsiderably affect the software development effort interms of productivity and diseconomy of scale ie foreach higher level of PMAT there is a decrease ineffort required increase in productivity rate andreduction in diseconomy of scale although the percentage of (increasedecrease) is not uniform amonglevels The results also demonstrate that this effect is
more considerable and significant for larger projects ascompared to smaller projects This necessitates theadoption of CMMI-based process improvement forsoftware organizations that are developing projects of
large sizes As a result of investigating the impact of process maturity on effort it seems reasonable tosupport the suggestion that PMAT is a significant inputto software cost estimation models
Future work in the area of CMMI-based processmaturity requires collecting historical data for each of
the 22 process areas used in CMMI in order to examinewhich process area has most impact on effort productivity and diseconomy of scale In additionunlike SW-CMM CMMI has two differentrepresentations Staged and Continuous This studyfocused on the staged representation therefore furtherresearch is recommended to be conducted on theimpact of CMMI-based Process capability from thecontinuous representation perspective
References
[1]
Alyahya A Ahmad R and Lee S ldquoImpact ofCMMI Based Software Process Maturity onCOCOMO IIs Effort Estimationrdquo International Arab Journal of Information Technology vol 7no 2 pp 129-137 2010
[2] Boehm B Clark B Horowitz E Westland CMadachy R and Selby R ldquoCost Models forFuture Software Life Cycle ProcessesCOCOMO 20rdquo in Proceedings of SpecialVolume on Software Process and Product Measurement Amsterdam pp 45-60 1995
[3] Boehm B Horowitz E Madachy R Reifer D
Clark B Steece B Brown A Chulani S andAbts C Software Cost Estimation withCOCOMO II Prentice Hall 2000
[4] Boehm B Software Engineering EconomicsPrentice Hall 1981
[5] Bollinger T and McGowan C ldquoA Critical Lookat Software Capability Evaluationsrdquo IEEESoftware vol 26 no 5 pp 80-83 2009
[6] Butler K ldquoThe Economic Benefits of SoftwareProcess Improvementrdquo in Proceedings of Crosstalk Hill AFB Ogden pp 14-17 1995
[7] Chrissis M Konrad M and Shrum S CMMI
Guidelines for Process Integration and Product Improvement 3ed Addison-Wesley 2003
[8] Clark B ldquoQuantifying the Effects of ProcessImprovement on Effortrdquo IEEE Softwarevol 17 no 6 pp 65-70 2000
[9] Clark B ldquoThe Effects of Software ProcessMaturity on Software Development Effortrdquo PhD Dissertation University of Southern California1997
[10] Damian D Zowghi D Vaidyanathasamy L andPal Y ldquoAn Industrial Experience in ProcessImprovement An Early Assessment at theAustralian Center for Unisys Softwarerdquo in Proceedings of International Symposium on Empirical Software Engineering (ISESE) pp111-123 2002
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144
983108 983145 983155 983141 983139 983151 983150 983151 983149 983161 983151 983140 983123 983139 983137 983148 983141
983120983154983151983139983141983155983155 983117983137983156983157983154983145983156983161 983116 983141983158983141983148
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983108983145983155983141983139983151983150983151983149983161 983151983140 983155983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 39
half and upper half According to [8] the CMM level 1(lower half) is for organizations that rely on ldquoheroesrdquoto do the job They do not focus on processes ordocumenting lessons learned The CMM level 1 (upperhalf) is for organizations that have implemented mostof the requirements that would satisfy CMM level 2 In
CMMrsquos published definition level 1 (lower half) and(upper half) are grouped into level 1
3 Research Motivation and Hypothesis
31 Research Motivation
An increasing number of software developmentorganizations around the world have adopted CMMI toimprove their software development processes [37] Inthe literature there are numerous studies onsignificance and benefits of increasing the CMM-basedorganizational maturity levels However it is observedthat there are very limited studies on the impact and benefits of the CMMI-based process maturity It seemsCMM is still receiving much attention than CMMIeven though there are continual and widespreaddemands on the evidence about the impact and benefitsof CMMI-based process maturity from organizationsthat are adopting it [10]
Furthermore most of the available studies andresearch which focused on CMMI are case studies based on quantitative data To the best of ourknowledge qualitative studies on the effect of
increasing the CMMI maturity levels are lacked in theliterature It is believed that CMMI with its highacceptance rate as an SPI framework requires a specialinvestigation on the impact and benefits of increasingits maturity levels
32 Research Hypothesis
The hypothesis of the work presented here is thatincreasing the level of CMMI-based process maturitywill results in the following
1 A decrease in the software development effort
2
An increase in the productivity rate3 A reduce the in diseconomy of scale
As stated earlier this study relies on the new CMMI- based PMAT rating values which are shown in Table2
Table 2 The new PMAT rating values
PMAT
Description
CMMI
Level 1
(Lower)
CMMI
Level 1
(Upper)
CMMI
Level 2
CMMI
Level 3
CMMI
Level 4
CMMI
Level 5
Rating
Levels
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
New PMATValues
755 571 381 208 103 000
Since COCOMO II considers CMM level 2 as thenominal rating for PMAT the same consideration is
for CMMI level 2 Thus the percentage of increase ordecrease in effort productivity and diseconomy ofscale will be compared against the nominal CMMI- based PMAT ratings
4
Related Literature
Numerous researches and case studies have shownmany benefits of enhancing organizational processmaturity by using different assessment approaches [622 23 27 36] Girish et al [18] conducted anempirical study to investigate the effects of CMM ontwo critical factors in Information Systems (IS)implementation strategy which are project performance and software quality They claimed thatCMM levels are associated with IS implementationstrategies and higher CMM levels are relate to higher project performance and software quality that lead to
noticeable reduction in software development effortand schedule From a review of seventeen publishedarticles Galin and Avrahami [15] explored CMM- based benefits such as defects rework schedule productivity error defection effectiveness and ReturnOn Investment (ROI) concluding that a goodinvestment in CMM programs leads to enhancedsoftware development and maintenance Diaz and King[11] claimed that increase in CMM process maturityresults in an improvement in quality phasecontainment productivity and rework In order toexplore the impact of process maturity on software
development effort and based on CMM with the aid of161-project sample Clark in [8] isolated the effects oneffort of process maturity versus effects of otherfactors and found that an increasing of oneorganizational process maturity level can result in areduction in software development effort by 4 to11 But this reduction seemed like a generalizationacross all five levels of CMM process maturity iethe percentage of effort reduction is not the sameamong all levels Memon has reported in [29] thatCMM maturity levels considerably influences thesoftware development effort and productivity El
Emam and Goldenson [13] in an comprehensivereview of studies and publications on theimplementation of SPI methodologies includingCMM reported qualitative performance improvementsin terms of higher quality higher productivityimproved ability to meet development schedulesDonald et al [12] conducted an empirical research tofind out the relationship between quality of the products organizational process maturity developmenteffort and project schedule Their findings indicatedthat process maturity has an effect in reducing softwaredevelopment schedule and effort Another survey- based study of individuals from SW-CMM-assessedsoftware organizations revealed that higher maturityorganizations are associated with better performanceincluding the ability to meet budget and schedule as
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 49
well as increase staff productivity product quality andcustomer satisfaction [20] Herbsleb and Goldenson in[21] showed solid evidence in a sample of 61 softwareorganizations that high CMM-based software processmaturity is associated with high performance
Despite the numerous studies that have investigated
the performance assessment results of CMM-basedsoftware process maturity and its impact on softwaredevelopment effort there are still very limited workson the overall benefits of CMMI-based software process maturity [37] Case studies have also shown benefits from CMMI-based software process maturityin [14 16 17 19 25 28 32 34 37] Goldenson andGibson in [19] reported some great and crediblequantitative evidence that CMMI-based software process maturity can help an organization achievehigher quality products and better project performancewith lower cost and decreased project effort Due to the
limitation of the performance results provided in [19]Gibson et al [17] continued the assessment performance of CMMI-based software processimprovement and provides empirical tangible evidenceabout the performance results that can achieve as anoutcome of CMMI-based process improvement Theyreported ldquoThere now is ample evidence that processimprovement using the CMMI Product Suite can resultin marked improvements in schedule and cost performance product quality return on investment andother measures of performance outcomerdquo
5
Research Method
In order to investigate the impact of CMMI-based process maturity level on a variety of software productsizes it is more recommended to classify the softwaresizes in an appropriate manner since size is consideredthe most influential factor in predicting effort of thesoftware product [9] Boehm in [4] has classified thesoftware product sizes as small intermediate mediumlarge and very large See Table 3
Table 3 Software size classification according to [4]
Classification Size (KLOC)
Small (S) 2
Intermediate (I) 8
Medium (M) 32
Large (L) 128
Very Large (VL) 512
The scale factor PMAT is used here to capture theimpact of different process maturity levels on softwaredevelopment effort for standard size projects classifiedabove
51 Effort Estimation
The basic idea in our research method is quantifyingthe impact of PMAT versus other factors that influence
the software development effort To do so we separatethe impact of PMAT from other factors because whendifferent kinds of improvements are carried outconcurrently in the organization project managers willhave no idea on how to determine the amount ofimprovement gained from process maturity with the
presence of other factors [8] In this contextCOCOMO II model is employed in order to estimatethe effort of software development Subsequently alleffort multipliers are set to be nominal ie the valueof 1 Furthermore as described in Table 4 all scalefactors except PMAT (the one related to the processmaturity) are also assumed to be nominal The effortmultipliers and scale factors (except PMAT) are set totheir nominal values in order to isolate their potentialeffects on software development effort As an examplefor a nominal PMAT rating and a standard large size project by substituting values in equations 1 and 2 we
get 294128 58521
Table 4 Nominal values for all scale factors in all rating levels(except CMMI-based PAMT)
Scale
Factor
(SF)
CMMI
Level 1
(lower)
CMMI
Level 1
(upper)
CMMI
Level 2
CMMI
Level 3
CMMI
Level 4
CMM I
Level 5
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
Precedentedness
(PREC)372 372 372 372 372 372
DevelopmentFlexibility (FLEX) 304 304 304 304 304 304
Risk
Resolution (RESL)424 424 424 424 424 424
Team
Cohesion
(TEAM)
329 329 329 329 329 329
New
Process
Maturity
(PMAT)
755 571 381 208 103 000
Summation
of All SF2184 2000 1810 1637 1532 1429
52 Productivity RateIn order to test our hypothesis which is increasing thelevel of CMMI-based process maturity increases the productivity rate equation 3 is applied for eachestimated effort
where SIZE is the standard size used and it ismeasured in this formula by thousand lines of codes(KLOC) and EFFORT is the effort estimated in eachPMAT level for all standard sizes
As an example for the productivity for nominalPMAT rating and standard large size project equation3 will be applied to the effort produced in the previoussection
(3)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 59
12858521 21872
53 Diseconomy of Scale
Diseconomy of scale refers to relatively more increasein effort as compared to the increase in size of a
software product That is doubling the project size willresult in more than double of the original project effort being needed to complete the project
To make this concept more clear Memon [29]suggested dividing the standard project sizes discussedearlier by the small size (2 KLOC) and then calledfrom small to intermediate (from S to I) from small tomedium (from S to M) from small to large (from S toL) and from small to very large (from S to VL)Similarly their corresponding efforts are also divided by the effort of small size in order to visualize theeffect of diseconomy of scale see Table 5
Table 5 Software sizes ratio
Classification Size Ration
From S to I 4From S to M 16From S to L 64
From S to VL 256
6 Evaluation Measure
In order to evaluate and compare our proposed resultsmeasures of effectiveness are required In this study
the percentage change in software development effort productivity rate and diseconomy of scale are used asthe primary Measures Of Effectiveness (MOE) The percent change means either increase or decrease ineffort productivity rate and diseconomy of scale Thiscriterion measure will determine the magnitude of theeffect of different process maturity levels ondevelopment effort for all standard sizes of software projects To compute these percentages we willassume the nominal rating of PMAT as the base caseHence the percent of change in effort productivityand diseconomy of scale is measured by using equation4
where Parameter refers to the computed value ofeffort productivity or diseconomy of scale at a particular PMAT rating Whereas Parameter nominal referto the nominal value of the same parameter
A combination of positive and negative changes will be seen in the resulted values A negative valueindicates percentage reduction in the parameter valueand a positive one show percentage increase in the
parameter value
7 Results and Discussions
71 Effort
After applying our method COCOMO II has estimatedthe effort for all standard project sizes in each PAMTrating Table 6 shows the resulted effort Table 7
shows the percentage change in effort in each processmaturity level for all standard size projects and Figure1 is a visual representation of Table 7
Table 6 Estimated effort in all CMMI-based PMAT ratings for allstandard sizes
Project Effort (PM) Based on PMAT Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 643 635 626 619 614 610
Intermediate 8 3072 2956 2842 2742 2682 2625
Medium 32 14681 13774 12896 12146 11712 11301
Large 128 70165 64173 58521 53809 51137 48644
Very Large 512 335342 298976 265559 238391 223277 209381
Table 7 Percent change of effort in all CMMI-based PMAT ratingsfor all standard sizes
Project Average Change in Effort
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 263 133 000 -119 -191 -261
Intermediate 8 809 403 000 -353 -562 -762
Medium 32 1384 681 000 -582 -919 -1237
Large 128 1990 966 000 -805 -1262 -1688
Very Large 512 2628 1258 000 -1023 -1592 -2115
Figure 1 Percent change of effort in all PMAT ratings for allstandard sizes
As seen in Table 6 the results indicate that for eachimprovement in the PMAT rating there is a decreasein the required software development effort It can benoticed that these improvements in the effort are moreconsiderable for larger projects size than for smallersizes Also the percentage change shown in Table 7
and graphically in Figure 1 obviously gives the sameindications The percent change in required effortvaries from increase of (263) for very low rating ofPMAT to a decrease of 261 for a rating of extra high(a total change of 524) for small size projects
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983109 983142 983142 983151 983154 983156
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983109983142983142983151983154983156 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141983154 983149983141 983140983145983137983156983141 983117983141983140983145983157983149 983116 983137983154983143983141 983126983141983154 983161 983116983137983154 983143983141
(4)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 69
whereas for very large projects the percent change inrequired effort varies from an increase of (2628) forvery low rating of PMAT to a decrease of 2115 for arating of extra high (a total change of 4743)
72 Productivity Rate
Table 8 and Figure 2 show the resulted productivityWhile the percent change in productivity in each process maturity level for all standard size projects areshown in Table 9 and Figure 3
Table 8 Productivity rate in all CMMI-based PMAT ratings for allstandard sizes
ProjectProductivity (KLOCPM) Based on ISF-PMAT
Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 311 315 31934 323 326 328
Intermediate 8 260 271 28150 292 298 305Medium 32 218 232 24813 263 273 283
Large 128 182 199 21872 238 250 263
Very Large 512 153 171 19280 215 229 245
Figure 2 Productivity rate in all PMAT ratings for all standardsizes
As can be seen in Table 8 and visually in Figure 2there is a considerable growth in the productivity ratein all PMAT levels They also give a clear indicationthat the productivity rate changes (increases) more
rapidly for larger projects as compared to smaller projects
Table 9 Percent change of productivity in all CMMI-based PMATratings for all standard sizes
Project Average Change in Productivity
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 -256 -131 000 121 195 268
Intermediate 8 -748 -387 000 366 595 824
Medium 32 -1216 -637 000 618 1011 1412
Large 128 -1660 -881 000 876 1444 2031
Very Large 512 -2081 -1118 000 1140 1894 2683
As shown in Table 9 and graphically in Figure 3 the percent change in productivity rate varies fromdecrease of (256) for very low rating of PMAT to an
increase of 268 for a rating of extra high (a totalchange of 524) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from a decrease of (2081) for very lowrating of PMAT to an increase of 2683 for a ratingof extra high (a total change of 4764)
Figure 3 Percent change of productivity in all PMAT ratings for allstandard sizes
73 Diseconomy of Scale
When the standard sizes were divided by the small size(2 KLOC) similarly their corresponding efforts arealso divided by the effort of small size to visualize theeffect of diseconomy of scale as shown in Table 10 andFigure 4 To visualize the effect of diseconomy ofscale the size ratio has been plotted in Figure 4 The percent change in diseconomy of scale in each process
maturity level for all standard size projects are shownin Table 11 and Figure 5
Table 10 Diseconomy of scale in all CMMI-based PMAT ratingsfor all standard sizes
ProjectDiseconomy of Scale Based on CMMI-Based
PMAT Ratings and Project Size
ClassificationSizeRatio
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 478 466 454 443 437 430
From S to M 16 2284 2171 2059 1963 1906 1853
From S to L 64 10917 10113 9344 8696 8324 7975
From S to VL 256 52174 47114 42402 38524 36345 34327
Figure 4 Diseconomy of scale in all PMAT ratings for all standardsizes
Table 10 and Figure 4 show that diseconomy ofscale improves desirably with improvement in PMAT
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983154983151983140983157983139983156983145983158983145983156983161 983122983137983156983141983155 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141 983154983149983141983140983145983137983156983141 983117983141 983140983145983157983149 983116 983137983154 983143983141 983126983141 983154983161 983116983137983154983143983141
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150 983137983148 983112983145 983143983144 983126983141 983154983161 983112983145 983143983144 983109983160 983156983154 983137 983112983145 983143983144
983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983120983154983151983140983157983139983156983145983158983145983156983161 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983141 983113983150983156983141983154983149983141 983140983145983137983156983141 983117983141 983140983145983157983149 983116983137983154983143983141 983126983141 983154983161 983116983137983154983143983141
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983093983088983088
983094983088983088
983123983145983162983141
983122983137983156983145983151
983126983141983154983161
983116983151983159
983116 983151983159 983118983151 983149983145983150 983137983148 983112983145983143983144 983126983141 983154983161
983112983145983143983144
983109983160983156983154983137
983112983145983143983144
983122 983137 983156 983145 983151 983151 983142 983109 983142 983142 983151 983154 983156 983087 983123 983145 983162 983141
983120983117983105983124 983148983141983158983141983148 983154983137983156983145983150983143 983087 983123983145983162983141 983080983115983116983119983107983081
983108983145983155983141983139983151983150983151983149983161 983151983142 983123983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 79
rating levels They also give an obvious andconsiderable indication that the productivity ratechanges (increases) more rapidly for larger projects ascompared to smaller projects
Table 11 Percent Change of Diseconomy of Scale in all CMMI- based PMAT Ratings for all Standard sizes
Project Average Change in Diseconomy of Scale
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 532 267 000 -237 -378 -514
From S to M 16 1093 541 000 -468 -742 -1002
From S to L 64 1683 822 000 -694 -1092 -1465
From S to VL 256 2305 1111 000 -915 -1429 -1904
Figure 5 Percent change of diseconomy of scale in all PMATratings for all standard sizes
As shown in Table 11 and graphically in Figure 5
the percent change in diseconomy of scale varies froman increase of (532) for very low rating of PMAT toa decrease of 514 for a rating of extra high (a totalchange of 1046) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from an increase of (2305) for very lowrating of PMAT to a decrease of 1904 for a rating ofextra high (a total change of 4209)
8 Conclusions and Future Work
This paper investigates the impact of CMMI-based
process maturity levels on software developmenteffort productivity rate and diseconomy of scale for aset of standard project sizes The result of thisinvestigation has boosted our hypotheses and revealedthat the process maturity rating levels based on CMMIconsiderably affect the software development effort interms of productivity and diseconomy of scale ie foreach higher level of PMAT there is a decrease ineffort required increase in productivity rate andreduction in diseconomy of scale although the percentage of (increasedecrease) is not uniform amonglevels The results also demonstrate that this effect is
more considerable and significant for larger projects ascompared to smaller projects This necessitates theadoption of CMMI-based process improvement forsoftware organizations that are developing projects of
large sizes As a result of investigating the impact of process maturity on effort it seems reasonable tosupport the suggestion that PMAT is a significant inputto software cost estimation models
Future work in the area of CMMI-based processmaturity requires collecting historical data for each of
the 22 process areas used in CMMI in order to examinewhich process area has most impact on effort productivity and diseconomy of scale In additionunlike SW-CMM CMMI has two differentrepresentations Staged and Continuous This studyfocused on the staged representation therefore furtherresearch is recommended to be conducted on theimpact of CMMI-based Process capability from thecontinuous representation perspective
References
[1]
Alyahya A Ahmad R and Lee S ldquoImpact ofCMMI Based Software Process Maturity onCOCOMO IIs Effort Estimationrdquo International Arab Journal of Information Technology vol 7no 2 pp 129-137 2010
[2] Boehm B Clark B Horowitz E Westland CMadachy R and Selby R ldquoCost Models forFuture Software Life Cycle ProcessesCOCOMO 20rdquo in Proceedings of SpecialVolume on Software Process and Product Measurement Amsterdam pp 45-60 1995
[3] Boehm B Horowitz E Madachy R Reifer D
Clark B Steece B Brown A Chulani S andAbts C Software Cost Estimation withCOCOMO II Prentice Hall 2000
[4] Boehm B Software Engineering EconomicsPrentice Hall 1981
[5] Bollinger T and McGowan C ldquoA Critical Lookat Software Capability Evaluationsrdquo IEEESoftware vol 26 no 5 pp 80-83 2009
[6] Butler K ldquoThe Economic Benefits of SoftwareProcess Improvementrdquo in Proceedings of Crosstalk Hill AFB Ogden pp 14-17 1995
[7] Chrissis M Konrad M and Shrum S CMMI
Guidelines for Process Integration and Product Improvement 3ed Addison-Wesley 2003
[8] Clark B ldquoQuantifying the Effects of ProcessImprovement on Effortrdquo IEEE Softwarevol 17 no 6 pp 65-70 2000
[9] Clark B ldquoThe Effects of Software ProcessMaturity on Software Development Effortrdquo PhD Dissertation University of Southern California1997
[10] Damian D Zowghi D Vaidyanathasamy L andPal Y ldquoAn Industrial Experience in ProcessImprovement An Early Assessment at theAustralian Center for Unisys Softwarerdquo in Proceedings of International Symposium on Empirical Software Engineering (ISESE) pp111-123 2002
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144
983108 983145 983155 983141 983139 983151 983150 983151 983149 983161 983151 983140 983123 983139 983137 983148 983141
983120983154983151983139983141983155983155 983117983137983156983157983154983145983156983161 983116 983141983158983141983148
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983108983145983155983141983139983151983150983151983149983161 983151983140 983155983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 49
well as increase staff productivity product quality andcustomer satisfaction [20] Herbsleb and Goldenson in[21] showed solid evidence in a sample of 61 softwareorganizations that high CMM-based software processmaturity is associated with high performance
Despite the numerous studies that have investigated
the performance assessment results of CMM-basedsoftware process maturity and its impact on softwaredevelopment effort there are still very limited workson the overall benefits of CMMI-based software process maturity [37] Case studies have also shown benefits from CMMI-based software process maturityin [14 16 17 19 25 28 32 34 37] Goldenson andGibson in [19] reported some great and crediblequantitative evidence that CMMI-based software process maturity can help an organization achievehigher quality products and better project performancewith lower cost and decreased project effort Due to the
limitation of the performance results provided in [19]Gibson et al [17] continued the assessment performance of CMMI-based software processimprovement and provides empirical tangible evidenceabout the performance results that can achieve as anoutcome of CMMI-based process improvement Theyreported ldquoThere now is ample evidence that processimprovement using the CMMI Product Suite can resultin marked improvements in schedule and cost performance product quality return on investment andother measures of performance outcomerdquo
5
Research Method
In order to investigate the impact of CMMI-based process maturity level on a variety of software productsizes it is more recommended to classify the softwaresizes in an appropriate manner since size is consideredthe most influential factor in predicting effort of thesoftware product [9] Boehm in [4] has classified thesoftware product sizes as small intermediate mediumlarge and very large See Table 3
Table 3 Software size classification according to [4]
Classification Size (KLOC)
Small (S) 2
Intermediate (I) 8
Medium (M) 32
Large (L) 128
Very Large (VL) 512
The scale factor PMAT is used here to capture theimpact of different process maturity levels on softwaredevelopment effort for standard size projects classifiedabove
51 Effort Estimation
The basic idea in our research method is quantifyingthe impact of PMAT versus other factors that influence
the software development effort To do so we separatethe impact of PMAT from other factors because whendifferent kinds of improvements are carried outconcurrently in the organization project managers willhave no idea on how to determine the amount ofimprovement gained from process maturity with the
presence of other factors [8] In this contextCOCOMO II model is employed in order to estimatethe effort of software development Subsequently alleffort multipliers are set to be nominal ie the valueof 1 Furthermore as described in Table 4 all scalefactors except PMAT (the one related to the processmaturity) are also assumed to be nominal The effortmultipliers and scale factors (except PMAT) are set totheir nominal values in order to isolate their potentialeffects on software development effort As an examplefor a nominal PMAT rating and a standard large size project by substituting values in equations 1 and 2 we
get 294128 58521
Table 4 Nominal values for all scale factors in all rating levels(except CMMI-based PAMT)
Scale
Factor
(SF)
CMMI
Level 1
(lower)
CMMI
Level 1
(upper)
CMMI
Level 2
CMMI
Level 3
CMMI
Level 4
CMM I
Level 5
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
Precedentedness
(PREC)372 372 372 372 372 372
DevelopmentFlexibility (FLEX) 304 304 304 304 304 304
Risk
Resolution (RESL)424 424 424 424 424 424
Team
Cohesion
(TEAM)
329 329 329 329 329 329
New
Process
Maturity
(PMAT)
755 571 381 208 103 000
Summation
of All SF2184 2000 1810 1637 1532 1429
52 Productivity RateIn order to test our hypothesis which is increasing thelevel of CMMI-based process maturity increases the productivity rate equation 3 is applied for eachestimated effort
where SIZE is the standard size used and it ismeasured in this formula by thousand lines of codes(KLOC) and EFFORT is the effort estimated in eachPMAT level for all standard sizes
As an example for the productivity for nominalPMAT rating and standard large size project equation3 will be applied to the effort produced in the previoussection
(3)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 59
12858521 21872
53 Diseconomy of Scale
Diseconomy of scale refers to relatively more increasein effort as compared to the increase in size of a
software product That is doubling the project size willresult in more than double of the original project effort being needed to complete the project
To make this concept more clear Memon [29]suggested dividing the standard project sizes discussedearlier by the small size (2 KLOC) and then calledfrom small to intermediate (from S to I) from small tomedium (from S to M) from small to large (from S toL) and from small to very large (from S to VL)Similarly their corresponding efforts are also divided by the effort of small size in order to visualize theeffect of diseconomy of scale see Table 5
Table 5 Software sizes ratio
Classification Size Ration
From S to I 4From S to M 16From S to L 64
From S to VL 256
6 Evaluation Measure
In order to evaluate and compare our proposed resultsmeasures of effectiveness are required In this study
the percentage change in software development effort productivity rate and diseconomy of scale are used asthe primary Measures Of Effectiveness (MOE) The percent change means either increase or decrease ineffort productivity rate and diseconomy of scale Thiscriterion measure will determine the magnitude of theeffect of different process maturity levels ondevelopment effort for all standard sizes of software projects To compute these percentages we willassume the nominal rating of PMAT as the base caseHence the percent of change in effort productivityand diseconomy of scale is measured by using equation4
where Parameter refers to the computed value ofeffort productivity or diseconomy of scale at a particular PMAT rating Whereas Parameter nominal referto the nominal value of the same parameter
A combination of positive and negative changes will be seen in the resulted values A negative valueindicates percentage reduction in the parameter valueand a positive one show percentage increase in the
parameter value
7 Results and Discussions
71 Effort
After applying our method COCOMO II has estimatedthe effort for all standard project sizes in each PAMTrating Table 6 shows the resulted effort Table 7
shows the percentage change in effort in each processmaturity level for all standard size projects and Figure1 is a visual representation of Table 7
Table 6 Estimated effort in all CMMI-based PMAT ratings for allstandard sizes
Project Effort (PM) Based on PMAT Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 643 635 626 619 614 610
Intermediate 8 3072 2956 2842 2742 2682 2625
Medium 32 14681 13774 12896 12146 11712 11301
Large 128 70165 64173 58521 53809 51137 48644
Very Large 512 335342 298976 265559 238391 223277 209381
Table 7 Percent change of effort in all CMMI-based PMAT ratingsfor all standard sizes
Project Average Change in Effort
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 263 133 000 -119 -191 -261
Intermediate 8 809 403 000 -353 -562 -762
Medium 32 1384 681 000 -582 -919 -1237
Large 128 1990 966 000 -805 -1262 -1688
Very Large 512 2628 1258 000 -1023 -1592 -2115
Figure 1 Percent change of effort in all PMAT ratings for allstandard sizes
As seen in Table 6 the results indicate that for eachimprovement in the PMAT rating there is a decreasein the required software development effort It can benoticed that these improvements in the effort are moreconsiderable for larger projects size than for smallersizes Also the percentage change shown in Table 7
and graphically in Figure 1 obviously gives the sameindications The percent change in required effortvaries from increase of (263) for very low rating ofPMAT to a decrease of 261 for a rating of extra high(a total change of 524) for small size projects
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983109 983142 983142 983151 983154 983156
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983109983142983142983151983154983156 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141983154 983149983141 983140983145983137983156983141 983117983141983140983145983157983149 983116 983137983154983143983141 983126983141983154 983161 983116983137983154 983143983141
(4)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 69
whereas for very large projects the percent change inrequired effort varies from an increase of (2628) forvery low rating of PMAT to a decrease of 2115 for arating of extra high (a total change of 4743)
72 Productivity Rate
Table 8 and Figure 2 show the resulted productivityWhile the percent change in productivity in each process maturity level for all standard size projects areshown in Table 9 and Figure 3
Table 8 Productivity rate in all CMMI-based PMAT ratings for allstandard sizes
ProjectProductivity (KLOCPM) Based on ISF-PMAT
Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 311 315 31934 323 326 328
Intermediate 8 260 271 28150 292 298 305Medium 32 218 232 24813 263 273 283
Large 128 182 199 21872 238 250 263
Very Large 512 153 171 19280 215 229 245
Figure 2 Productivity rate in all PMAT ratings for all standardsizes
As can be seen in Table 8 and visually in Figure 2there is a considerable growth in the productivity ratein all PMAT levels They also give a clear indicationthat the productivity rate changes (increases) more
rapidly for larger projects as compared to smaller projects
Table 9 Percent change of productivity in all CMMI-based PMATratings for all standard sizes
Project Average Change in Productivity
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 -256 -131 000 121 195 268
Intermediate 8 -748 -387 000 366 595 824
Medium 32 -1216 -637 000 618 1011 1412
Large 128 -1660 -881 000 876 1444 2031
Very Large 512 -2081 -1118 000 1140 1894 2683
As shown in Table 9 and graphically in Figure 3 the percent change in productivity rate varies fromdecrease of (256) for very low rating of PMAT to an
increase of 268 for a rating of extra high (a totalchange of 524) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from a decrease of (2081) for very lowrating of PMAT to an increase of 2683 for a ratingof extra high (a total change of 4764)
Figure 3 Percent change of productivity in all PMAT ratings for allstandard sizes
73 Diseconomy of Scale
When the standard sizes were divided by the small size(2 KLOC) similarly their corresponding efforts arealso divided by the effort of small size to visualize theeffect of diseconomy of scale as shown in Table 10 andFigure 4 To visualize the effect of diseconomy ofscale the size ratio has been plotted in Figure 4 The percent change in diseconomy of scale in each process
maturity level for all standard size projects are shownin Table 11 and Figure 5
Table 10 Diseconomy of scale in all CMMI-based PMAT ratingsfor all standard sizes
ProjectDiseconomy of Scale Based on CMMI-Based
PMAT Ratings and Project Size
ClassificationSizeRatio
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 478 466 454 443 437 430
From S to M 16 2284 2171 2059 1963 1906 1853
From S to L 64 10917 10113 9344 8696 8324 7975
From S to VL 256 52174 47114 42402 38524 36345 34327
Figure 4 Diseconomy of scale in all PMAT ratings for all standardsizes
Table 10 and Figure 4 show that diseconomy ofscale improves desirably with improvement in PMAT
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983154983151983140983157983139983156983145983158983145983156983161 983122983137983156983141983155 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141 983154983149983141983140983145983137983156983141 983117983141 983140983145983157983149 983116 983137983154 983143983141 983126983141 983154983161 983116983137983154983143983141
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150 983137983148 983112983145 983143983144 983126983141 983154983161 983112983145 983143983144 983109983160 983156983154 983137 983112983145 983143983144
983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983120983154983151983140983157983139983156983145983158983145983156983161 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983141 983113983150983156983141983154983149983141 983140983145983137983156983141 983117983141 983140983145983157983149 983116983137983154983143983141 983126983141 983154983161 983116983137983154983143983141
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983093983088983088
983094983088983088
983123983145983162983141
983122983137983156983145983151
983126983141983154983161
983116983151983159
983116 983151983159 983118983151 983149983145983150 983137983148 983112983145983143983144 983126983141 983154983161
983112983145983143983144
983109983160983156983154983137
983112983145983143983144
983122 983137 983156 983145 983151 983151 983142 983109 983142 983142 983151 983154 983156 983087 983123 983145 983162 983141
983120983117983105983124 983148983141983158983141983148 983154983137983156983145983150983143 983087 983123983145983162983141 983080983115983116983119983107983081
983108983145983155983141983139983151983150983151983149983161 983151983142 983123983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 79
rating levels They also give an obvious andconsiderable indication that the productivity ratechanges (increases) more rapidly for larger projects ascompared to smaller projects
Table 11 Percent Change of Diseconomy of Scale in all CMMI- based PMAT Ratings for all Standard sizes
Project Average Change in Diseconomy of Scale
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 532 267 000 -237 -378 -514
From S to M 16 1093 541 000 -468 -742 -1002
From S to L 64 1683 822 000 -694 -1092 -1465
From S to VL 256 2305 1111 000 -915 -1429 -1904
Figure 5 Percent change of diseconomy of scale in all PMATratings for all standard sizes
As shown in Table 11 and graphically in Figure 5
the percent change in diseconomy of scale varies froman increase of (532) for very low rating of PMAT toa decrease of 514 for a rating of extra high (a totalchange of 1046) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from an increase of (2305) for very lowrating of PMAT to a decrease of 1904 for a rating ofextra high (a total change of 4209)
8 Conclusions and Future Work
This paper investigates the impact of CMMI-based
process maturity levels on software developmenteffort productivity rate and diseconomy of scale for aset of standard project sizes The result of thisinvestigation has boosted our hypotheses and revealedthat the process maturity rating levels based on CMMIconsiderably affect the software development effort interms of productivity and diseconomy of scale ie foreach higher level of PMAT there is a decrease ineffort required increase in productivity rate andreduction in diseconomy of scale although the percentage of (increasedecrease) is not uniform amonglevels The results also demonstrate that this effect is
more considerable and significant for larger projects ascompared to smaller projects This necessitates theadoption of CMMI-based process improvement forsoftware organizations that are developing projects of
large sizes As a result of investigating the impact of process maturity on effort it seems reasonable tosupport the suggestion that PMAT is a significant inputto software cost estimation models
Future work in the area of CMMI-based processmaturity requires collecting historical data for each of
the 22 process areas used in CMMI in order to examinewhich process area has most impact on effort productivity and diseconomy of scale In additionunlike SW-CMM CMMI has two differentrepresentations Staged and Continuous This studyfocused on the staged representation therefore furtherresearch is recommended to be conducted on theimpact of CMMI-based Process capability from thecontinuous representation perspective
References
[1]
Alyahya A Ahmad R and Lee S ldquoImpact ofCMMI Based Software Process Maturity onCOCOMO IIs Effort Estimationrdquo International Arab Journal of Information Technology vol 7no 2 pp 129-137 2010
[2] Boehm B Clark B Horowitz E Westland CMadachy R and Selby R ldquoCost Models forFuture Software Life Cycle ProcessesCOCOMO 20rdquo in Proceedings of SpecialVolume on Software Process and Product Measurement Amsterdam pp 45-60 1995
[3] Boehm B Horowitz E Madachy R Reifer D
Clark B Steece B Brown A Chulani S andAbts C Software Cost Estimation withCOCOMO II Prentice Hall 2000
[4] Boehm B Software Engineering EconomicsPrentice Hall 1981
[5] Bollinger T and McGowan C ldquoA Critical Lookat Software Capability Evaluationsrdquo IEEESoftware vol 26 no 5 pp 80-83 2009
[6] Butler K ldquoThe Economic Benefits of SoftwareProcess Improvementrdquo in Proceedings of Crosstalk Hill AFB Ogden pp 14-17 1995
[7] Chrissis M Konrad M and Shrum S CMMI
Guidelines for Process Integration and Product Improvement 3ed Addison-Wesley 2003
[8] Clark B ldquoQuantifying the Effects of ProcessImprovement on Effortrdquo IEEE Softwarevol 17 no 6 pp 65-70 2000
[9] Clark B ldquoThe Effects of Software ProcessMaturity on Software Development Effortrdquo PhD Dissertation University of Southern California1997
[10] Damian D Zowghi D Vaidyanathasamy L andPal Y ldquoAn Industrial Experience in ProcessImprovement An Early Assessment at theAustralian Center for Unisys Softwarerdquo in Proceedings of International Symposium on Empirical Software Engineering (ISESE) pp111-123 2002
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144
983108 983145 983155 983141 983139 983151 983150 983151 983149 983161 983151 983140 983123 983139 983137 983148 983141
983120983154983151983139983141983155983155 983117983137983156983157983154983145983156983161 983116 983141983158983141983148
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983108983145983155983141983139983151983150983151983149983161 983151983140 983155983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 59
12858521 21872
53 Diseconomy of Scale
Diseconomy of scale refers to relatively more increasein effort as compared to the increase in size of a
software product That is doubling the project size willresult in more than double of the original project effort being needed to complete the project
To make this concept more clear Memon [29]suggested dividing the standard project sizes discussedearlier by the small size (2 KLOC) and then calledfrom small to intermediate (from S to I) from small tomedium (from S to M) from small to large (from S toL) and from small to very large (from S to VL)Similarly their corresponding efforts are also divided by the effort of small size in order to visualize theeffect of diseconomy of scale see Table 5
Table 5 Software sizes ratio
Classification Size Ration
From S to I 4From S to M 16From S to L 64
From S to VL 256
6 Evaluation Measure
In order to evaluate and compare our proposed resultsmeasures of effectiveness are required In this study
the percentage change in software development effort productivity rate and diseconomy of scale are used asthe primary Measures Of Effectiveness (MOE) The percent change means either increase or decrease ineffort productivity rate and diseconomy of scale Thiscriterion measure will determine the magnitude of theeffect of different process maturity levels ondevelopment effort for all standard sizes of software projects To compute these percentages we willassume the nominal rating of PMAT as the base caseHence the percent of change in effort productivityand diseconomy of scale is measured by using equation4
where Parameter refers to the computed value ofeffort productivity or diseconomy of scale at a particular PMAT rating Whereas Parameter nominal referto the nominal value of the same parameter
A combination of positive and negative changes will be seen in the resulted values A negative valueindicates percentage reduction in the parameter valueand a positive one show percentage increase in the
parameter value
7 Results and Discussions
71 Effort
After applying our method COCOMO II has estimatedthe effort for all standard project sizes in each PAMTrating Table 6 shows the resulted effort Table 7
shows the percentage change in effort in each processmaturity level for all standard size projects and Figure1 is a visual representation of Table 7
Table 6 Estimated effort in all CMMI-based PMAT ratings for allstandard sizes
Project Effort (PM) Based on PMAT Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 643 635 626 619 614 610
Intermediate 8 3072 2956 2842 2742 2682 2625
Medium 32 14681 13774 12896 12146 11712 11301
Large 128 70165 64173 58521 53809 51137 48644
Very Large 512 335342 298976 265559 238391 223277 209381
Table 7 Percent change of effort in all CMMI-based PMAT ratingsfor all standard sizes
Project Average Change in Effort
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 263 133 000 -119 -191 -261
Intermediate 8 809 403 000 -353 -562 -762
Medium 32 1384 681 000 -582 -919 -1237
Large 128 1990 966 000 -805 -1262 -1688
Very Large 512 2628 1258 000 -1023 -1592 -2115
Figure 1 Percent change of effort in all PMAT ratings for allstandard sizes
As seen in Table 6 the results indicate that for eachimprovement in the PMAT rating there is a decreasein the required software development effort It can benoticed that these improvements in the effort are moreconsiderable for larger projects size than for smallersizes Also the percentage change shown in Table 7
and graphically in Figure 1 obviously gives the sameindications The percent change in required effortvaries from increase of (263) for very low rating ofPMAT to a decrease of 261 for a rating of extra high(a total change of 524) for small size projects
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983109 983142 983142 983151 983154 983156
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983109983142983142983151983154983156 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141983154 983149983141 983140983145983137983156983141 983117983141983140983145983157983149 983116 983137983154983143983141 983126983141983154 983161 983116983137983154 983143983141
(4)
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 69
whereas for very large projects the percent change inrequired effort varies from an increase of (2628) forvery low rating of PMAT to a decrease of 2115 for arating of extra high (a total change of 4743)
72 Productivity Rate
Table 8 and Figure 2 show the resulted productivityWhile the percent change in productivity in each process maturity level for all standard size projects areshown in Table 9 and Figure 3
Table 8 Productivity rate in all CMMI-based PMAT ratings for allstandard sizes
ProjectProductivity (KLOCPM) Based on ISF-PMAT
Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 311 315 31934 323 326 328
Intermediate 8 260 271 28150 292 298 305Medium 32 218 232 24813 263 273 283
Large 128 182 199 21872 238 250 263
Very Large 512 153 171 19280 215 229 245
Figure 2 Productivity rate in all PMAT ratings for all standardsizes
As can be seen in Table 8 and visually in Figure 2there is a considerable growth in the productivity ratein all PMAT levels They also give a clear indicationthat the productivity rate changes (increases) more
rapidly for larger projects as compared to smaller projects
Table 9 Percent change of productivity in all CMMI-based PMATratings for all standard sizes
Project Average Change in Productivity
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 -256 -131 000 121 195 268
Intermediate 8 -748 -387 000 366 595 824
Medium 32 -1216 -637 000 618 1011 1412
Large 128 -1660 -881 000 876 1444 2031
Very Large 512 -2081 -1118 000 1140 1894 2683
As shown in Table 9 and graphically in Figure 3 the percent change in productivity rate varies fromdecrease of (256) for very low rating of PMAT to an
increase of 268 for a rating of extra high (a totalchange of 524) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from a decrease of (2081) for very lowrating of PMAT to an increase of 2683 for a ratingof extra high (a total change of 4764)
Figure 3 Percent change of productivity in all PMAT ratings for allstandard sizes
73 Diseconomy of Scale
When the standard sizes were divided by the small size(2 KLOC) similarly their corresponding efforts arealso divided by the effort of small size to visualize theeffect of diseconomy of scale as shown in Table 10 andFigure 4 To visualize the effect of diseconomy ofscale the size ratio has been plotted in Figure 4 The percent change in diseconomy of scale in each process
maturity level for all standard size projects are shownin Table 11 and Figure 5
Table 10 Diseconomy of scale in all CMMI-based PMAT ratingsfor all standard sizes
ProjectDiseconomy of Scale Based on CMMI-Based
PMAT Ratings and Project Size
ClassificationSizeRatio
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 478 466 454 443 437 430
From S to M 16 2284 2171 2059 1963 1906 1853
From S to L 64 10917 10113 9344 8696 8324 7975
From S to VL 256 52174 47114 42402 38524 36345 34327
Figure 4 Diseconomy of scale in all PMAT ratings for all standardsizes
Table 10 and Figure 4 show that diseconomy ofscale improves desirably with improvement in PMAT
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983154983151983140983157983139983156983145983158983145983156983161 983122983137983156983141983155 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141 983154983149983141983140983145983137983156983141 983117983141 983140983145983157983149 983116 983137983154 983143983141 983126983141 983154983161 983116983137983154983143983141
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150 983137983148 983112983145 983143983144 983126983141 983154983161 983112983145 983143983144 983109983160 983156983154 983137 983112983145 983143983144
983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983120983154983151983140983157983139983156983145983158983145983156983161 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983141 983113983150983156983141983154983149983141 983140983145983137983156983141 983117983141 983140983145983157983149 983116983137983154983143983141 983126983141 983154983161 983116983137983154983143983141
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983093983088983088
983094983088983088
983123983145983162983141
983122983137983156983145983151
983126983141983154983161
983116983151983159
983116 983151983159 983118983151 983149983145983150 983137983148 983112983145983143983144 983126983141 983154983161
983112983145983143983144
983109983160983156983154983137
983112983145983143983144
983122 983137 983156 983145 983151 983151 983142 983109 983142 983142 983151 983154 983156 983087 983123 983145 983162 983141
983120983117983105983124 983148983141983158983141983148 983154983137983156983145983150983143 983087 983123983145983162983141 983080983115983116983119983107983081
983108983145983155983141983139983151983150983151983149983161 983151983142 983123983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 79
rating levels They also give an obvious andconsiderable indication that the productivity ratechanges (increases) more rapidly for larger projects ascompared to smaller projects
Table 11 Percent Change of Diseconomy of Scale in all CMMI- based PMAT Ratings for all Standard sizes
Project Average Change in Diseconomy of Scale
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 532 267 000 -237 -378 -514
From S to M 16 1093 541 000 -468 -742 -1002
From S to L 64 1683 822 000 -694 -1092 -1465
From S to VL 256 2305 1111 000 -915 -1429 -1904
Figure 5 Percent change of diseconomy of scale in all PMATratings for all standard sizes
As shown in Table 11 and graphically in Figure 5
the percent change in diseconomy of scale varies froman increase of (532) for very low rating of PMAT toa decrease of 514 for a rating of extra high (a totalchange of 1046) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from an increase of (2305) for very lowrating of PMAT to a decrease of 1904 for a rating ofextra high (a total change of 4209)
8 Conclusions and Future Work
This paper investigates the impact of CMMI-based
process maturity levels on software developmenteffort productivity rate and diseconomy of scale for aset of standard project sizes The result of thisinvestigation has boosted our hypotheses and revealedthat the process maturity rating levels based on CMMIconsiderably affect the software development effort interms of productivity and diseconomy of scale ie foreach higher level of PMAT there is a decrease ineffort required increase in productivity rate andreduction in diseconomy of scale although the percentage of (increasedecrease) is not uniform amonglevels The results also demonstrate that this effect is
more considerable and significant for larger projects ascompared to smaller projects This necessitates theadoption of CMMI-based process improvement forsoftware organizations that are developing projects of
large sizes As a result of investigating the impact of process maturity on effort it seems reasonable tosupport the suggestion that PMAT is a significant inputto software cost estimation models
Future work in the area of CMMI-based processmaturity requires collecting historical data for each of
the 22 process areas used in CMMI in order to examinewhich process area has most impact on effort productivity and diseconomy of scale In additionunlike SW-CMM CMMI has two differentrepresentations Staged and Continuous This studyfocused on the staged representation therefore furtherresearch is recommended to be conducted on theimpact of CMMI-based Process capability from thecontinuous representation perspective
References
[1]
Alyahya A Ahmad R and Lee S ldquoImpact ofCMMI Based Software Process Maturity onCOCOMO IIs Effort Estimationrdquo International Arab Journal of Information Technology vol 7no 2 pp 129-137 2010
[2] Boehm B Clark B Horowitz E Westland CMadachy R and Selby R ldquoCost Models forFuture Software Life Cycle ProcessesCOCOMO 20rdquo in Proceedings of SpecialVolume on Software Process and Product Measurement Amsterdam pp 45-60 1995
[3] Boehm B Horowitz E Madachy R Reifer D
Clark B Steece B Brown A Chulani S andAbts C Software Cost Estimation withCOCOMO II Prentice Hall 2000
[4] Boehm B Software Engineering EconomicsPrentice Hall 1981
[5] Bollinger T and McGowan C ldquoA Critical Lookat Software Capability Evaluationsrdquo IEEESoftware vol 26 no 5 pp 80-83 2009
[6] Butler K ldquoThe Economic Benefits of SoftwareProcess Improvementrdquo in Proceedings of Crosstalk Hill AFB Ogden pp 14-17 1995
[7] Chrissis M Konrad M and Shrum S CMMI
Guidelines for Process Integration and Product Improvement 3ed Addison-Wesley 2003
[8] Clark B ldquoQuantifying the Effects of ProcessImprovement on Effortrdquo IEEE Softwarevol 17 no 6 pp 65-70 2000
[9] Clark B ldquoThe Effects of Software ProcessMaturity on Software Development Effortrdquo PhD Dissertation University of Southern California1997
[10] Damian D Zowghi D Vaidyanathasamy L andPal Y ldquoAn Industrial Experience in ProcessImprovement An Early Assessment at theAustralian Center for Unisys Softwarerdquo in Proceedings of International Symposium on Empirical Software Engineering (ISESE) pp111-123 2002
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144
983108 983145 983155 983141 983139 983151 983150 983151 983149 983161 983151 983140 983123 983139 983137 983148 983141
983120983154983151983139983141983155983155 983117983137983156983157983154983145983156983161 983116 983141983158983141983148
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983108983145983155983141983139983151983150983151983149983161 983151983140 983155983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 69
whereas for very large projects the percent change inrequired effort varies from an increase of (2628) forvery low rating of PMAT to a decrease of 2115 for arating of extra high (a total change of 4743)
72 Productivity Rate
Table 8 and Figure 2 show the resulted productivityWhile the percent change in productivity in each process maturity level for all standard size projects areshown in Table 9 and Figure 3
Table 8 Productivity rate in all CMMI-based PMAT ratings for allstandard sizes
ProjectProductivity (KLOCPM) Based on ISF-PMAT
Ratings and Project Size
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 311 315 31934 323 326 328
Intermediate 8 260 271 28150 292 298 305Medium 32 218 232 24813 263 273 283
Large 128 182 199 21872 238 250 263
Very Large 512 153 171 19280 215 229 245
Figure 2 Productivity rate in all PMAT ratings for all standardsizes
As can be seen in Table 8 and visually in Figure 2there is a considerable growth in the productivity ratein all PMAT levels They also give a clear indicationthat the productivity rate changes (increases) more
rapidly for larger projects as compared to smaller projects
Table 9 Percent change of productivity in all CMMI-based PMATratings for all standard sizes
Project Average Change in Productivity
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
Small 2 -256 -131 000 121 195 268
Intermediate 8 -748 -387 000 366 595 824
Medium 32 -1216 -637 000 618 1011 1412
Large 128 -1660 -881 000 876 1444 2031
Very Large 512 -2081 -1118 000 1140 1894 2683
As shown in Table 9 and graphically in Figure 3 the percent change in productivity rate varies fromdecrease of (256) for very low rating of PMAT to an
increase of 268 for a rating of extra high (a totalchange of 524) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from a decrease of (2081) for very lowrating of PMAT to an increase of 2683 for a ratingof extra high (a total change of 4764)
Figure 3 Percent change of productivity in all PMAT ratings for allstandard sizes
73 Diseconomy of Scale
When the standard sizes were divided by the small size(2 KLOC) similarly their corresponding efforts arealso divided by the effort of small size to visualize theeffect of diseconomy of scale as shown in Table 10 andFigure 4 To visualize the effect of diseconomy ofscale the size ratio has been plotted in Figure 4 The percent change in diseconomy of scale in each process
maturity level for all standard size projects are shownin Table 11 and Figure 5
Table 10 Diseconomy of scale in all CMMI-based PMAT ratingsfor all standard sizes
ProjectDiseconomy of Scale Based on CMMI-Based
PMAT Ratings and Project Size
ClassificationSizeRatio
VeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 478 466 454 443 437 430
From S to M 16 2284 2171 2059 1963 1906 1853
From S to L 64 10917 10113 9344 8696 8324 7975
From S to VL 256 52174 47114 42402 38524 36345 34327
Figure 4 Diseconomy of scale in all PMAT ratings for all standardsizes
Table 10 and Figure 4 show that diseconomy ofscale improves desirably with improvement in PMAT
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983126983141983154983161 983116983151983159 983116983151983159 983118983151983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144 983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983154983151983140983157983139983156983145983158983145983156983161 983122983137983156983141983155 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983148 983113983150983156983141 983154983149983141983140983145983137983156983141 983117983141 983140983145983157983149 983116 983137983154 983143983141 983126983141 983154983161 983116983137983154983143983141
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150 983137983148 983112983145 983143983144 983126983141 983154983161 983112983145 983143983144 983109983160 983156983154 983137 983112983145 983143983144
983120 983154 983151 983140 983157 983139 983156 983145 983158 983145 983156 983161 983080 983115 983116 983119 983107 983087 983120 983117 983081
983120983117983105983124 983122983137983156983145983150983143983155
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983120983154983151983140983157983139983156983145983158983145983156983161 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983123983149983137983148983141 983113983150983156983141983154983149983141 983140983145983137983156983141 983117983141 983140983145983157983149 983116983137983154983143983141 983126983141 983154983161 983116983137983154983143983141
983088
983089983088983088
983090983088983088
983091983088983088
983092983088983088
983093983088983088
983094983088983088
983123983145983162983141
983122983137983156983145983151
983126983141983154983161
983116983151983159
983116 983151983159 983118983151 983149983145983150 983137983148 983112983145983143983144 983126983141 983154983161
983112983145983143983144
983109983160983156983154983137
983112983145983143983144
983122 983137 983156 983145 983151 983151 983142 983109 983142 983142 983151 983154 983156 983087 983123 983145 983162 983141
983120983117983105983124 983148983141983158983141983148 983154983137983156983145983150983143 983087 983123983145983162983141 983080983115983116983119983107983081
983108983145983155983141983139983151983150983151983149983161 983151983142 983123983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 79
rating levels They also give an obvious andconsiderable indication that the productivity ratechanges (increases) more rapidly for larger projects ascompared to smaller projects
Table 11 Percent Change of Diseconomy of Scale in all CMMI- based PMAT Ratings for all Standard sizes
Project Average Change in Diseconomy of Scale
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 532 267 000 -237 -378 -514
From S to M 16 1093 541 000 -468 -742 -1002
From S to L 64 1683 822 000 -694 -1092 -1465
From S to VL 256 2305 1111 000 -915 -1429 -1904
Figure 5 Percent change of diseconomy of scale in all PMATratings for all standard sizes
As shown in Table 11 and graphically in Figure 5
the percent change in diseconomy of scale varies froman increase of (532) for very low rating of PMAT toa decrease of 514 for a rating of extra high (a totalchange of 1046) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from an increase of (2305) for very lowrating of PMAT to a decrease of 1904 for a rating ofextra high (a total change of 4209)
8 Conclusions and Future Work
This paper investigates the impact of CMMI-based
process maturity levels on software developmenteffort productivity rate and diseconomy of scale for aset of standard project sizes The result of thisinvestigation has boosted our hypotheses and revealedthat the process maturity rating levels based on CMMIconsiderably affect the software development effort interms of productivity and diseconomy of scale ie foreach higher level of PMAT there is a decrease ineffort required increase in productivity rate andreduction in diseconomy of scale although the percentage of (increasedecrease) is not uniform amonglevels The results also demonstrate that this effect is
more considerable and significant for larger projects ascompared to smaller projects This necessitates theadoption of CMMI-based process improvement forsoftware organizations that are developing projects of
large sizes As a result of investigating the impact of process maturity on effort it seems reasonable tosupport the suggestion that PMAT is a significant inputto software cost estimation models
Future work in the area of CMMI-based processmaturity requires collecting historical data for each of
the 22 process areas used in CMMI in order to examinewhich process area has most impact on effort productivity and diseconomy of scale In additionunlike SW-CMM CMMI has two differentrepresentations Staged and Continuous This studyfocused on the staged representation therefore furtherresearch is recommended to be conducted on theimpact of CMMI-based Process capability from thecontinuous representation perspective
References
[1]
Alyahya A Ahmad R and Lee S ldquoImpact ofCMMI Based Software Process Maturity onCOCOMO IIs Effort Estimationrdquo International Arab Journal of Information Technology vol 7no 2 pp 129-137 2010
[2] Boehm B Clark B Horowitz E Westland CMadachy R and Selby R ldquoCost Models forFuture Software Life Cycle ProcessesCOCOMO 20rdquo in Proceedings of SpecialVolume on Software Process and Product Measurement Amsterdam pp 45-60 1995
[3] Boehm B Horowitz E Madachy R Reifer D
Clark B Steece B Brown A Chulani S andAbts C Software Cost Estimation withCOCOMO II Prentice Hall 2000
[4] Boehm B Software Engineering EconomicsPrentice Hall 1981
[5] Bollinger T and McGowan C ldquoA Critical Lookat Software Capability Evaluationsrdquo IEEESoftware vol 26 no 5 pp 80-83 2009
[6] Butler K ldquoThe Economic Benefits of SoftwareProcess Improvementrdquo in Proceedings of Crosstalk Hill AFB Ogden pp 14-17 1995
[7] Chrissis M Konrad M and Shrum S CMMI
Guidelines for Process Integration and Product Improvement 3ed Addison-Wesley 2003
[8] Clark B ldquoQuantifying the Effects of ProcessImprovement on Effortrdquo IEEE Softwarevol 17 no 6 pp 65-70 2000
[9] Clark B ldquoThe Effects of Software ProcessMaturity on Software Development Effortrdquo PhD Dissertation University of Southern California1997
[10] Damian D Zowghi D Vaidyanathasamy L andPal Y ldquoAn Industrial Experience in ProcessImprovement An Early Assessment at theAustralian Center for Unisys Softwarerdquo in Proceedings of International Symposium on Empirical Software Engineering (ISESE) pp111-123 2002
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144
983108 983145 983155 983141 983139 983151 983150 983151 983149 983161 983151 983140 983123 983139 983137 983148 983141
983120983154983151983139983141983155983155 983117983137983156983157983154983145983156983161 983116 983141983158983141983148
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983108983145983155983141983139983151983150983151983149983161 983151983140 983155983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 79
rating levels They also give an obvious andconsiderable indication that the productivity ratechanges (increases) more rapidly for larger projects ascompared to smaller projects
Table 11 Percent Change of Diseconomy of Scale in all CMMI- based PMAT Ratings for all Standard sizes
Project Average Change in Diseconomy of Scale
Classification SizeVeryLow
Low Nominal HighVeryHigh
ExtraHigh
From S to I 4 532 267 000 -237 -378 -514
From S to M 16 1093 541 000 -468 -742 -1002
From S to L 64 1683 822 000 -694 -1092 -1465
From S to VL 256 2305 1111 000 -915 -1429 -1904
Figure 5 Percent change of diseconomy of scale in all PMATratings for all standard sizes
As shown in Table 11 and graphically in Figure 5
the percent change in diseconomy of scale varies froman increase of (532) for very low rating of PMAT toa decrease of 514 for a rating of extra high (a totalchange of 1046) for small size projects Whereas forvery large projects the percent change in requiredeffort varies from an increase of (2305) for very lowrating of PMAT to a decrease of 1904 for a rating ofextra high (a total change of 4209)
8 Conclusions and Future Work
This paper investigates the impact of CMMI-based
process maturity levels on software developmenteffort productivity rate and diseconomy of scale for aset of standard project sizes The result of thisinvestigation has boosted our hypotheses and revealedthat the process maturity rating levels based on CMMIconsiderably affect the software development effort interms of productivity and diseconomy of scale ie foreach higher level of PMAT there is a decrease ineffort required increase in productivity rate andreduction in diseconomy of scale although the percentage of (increasedecrease) is not uniform amonglevels The results also demonstrate that this effect is
more considerable and significant for larger projects ascompared to smaller projects This necessitates theadoption of CMMI-based process improvement forsoftware organizations that are developing projects of
large sizes As a result of investigating the impact of process maturity on effort it seems reasonable tosupport the suggestion that PMAT is a significant inputto software cost estimation models
Future work in the area of CMMI-based processmaturity requires collecting historical data for each of
the 22 process areas used in CMMI in order to examinewhich process area has most impact on effort productivity and diseconomy of scale In additionunlike SW-CMM CMMI has two differentrepresentations Staged and Continuous This studyfocused on the staged representation therefore furtherresearch is recommended to be conducted on theimpact of CMMI-based Process capability from thecontinuous representation perspective
References
[1]
Alyahya A Ahmad R and Lee S ldquoImpact ofCMMI Based Software Process Maturity onCOCOMO IIs Effort Estimationrdquo International Arab Journal of Information Technology vol 7no 2 pp 129-137 2010
[2] Boehm B Clark B Horowitz E Westland CMadachy R and Selby R ldquoCost Models forFuture Software Life Cycle ProcessesCOCOMO 20rdquo in Proceedings of SpecialVolume on Software Process and Product Measurement Amsterdam pp 45-60 1995
[3] Boehm B Horowitz E Madachy R Reifer D
Clark B Steece B Brown A Chulani S andAbts C Software Cost Estimation withCOCOMO II Prentice Hall 2000
[4] Boehm B Software Engineering EconomicsPrentice Hall 1981
[5] Bollinger T and McGowan C ldquoA Critical Lookat Software Capability Evaluationsrdquo IEEESoftware vol 26 no 5 pp 80-83 2009
[6] Butler K ldquoThe Economic Benefits of SoftwareProcess Improvementrdquo in Proceedings of Crosstalk Hill AFB Ogden pp 14-17 1995
[7] Chrissis M Konrad M and Shrum S CMMI
Guidelines for Process Integration and Product Improvement 3ed Addison-Wesley 2003
[8] Clark B ldquoQuantifying the Effects of ProcessImprovement on Effortrdquo IEEE Softwarevol 17 no 6 pp 65-70 2000
[9] Clark B ldquoThe Effects of Software ProcessMaturity on Software Development Effortrdquo PhD Dissertation University of Southern California1997
[10] Damian D Zowghi D Vaidyanathasamy L andPal Y ldquoAn Industrial Experience in ProcessImprovement An Early Assessment at theAustralian Center for Unisys Softwarerdquo in Proceedings of International Symposium on Empirical Software Engineering (ISESE) pp111-123 2002
983085983091983088983086983088983088
983085983090983088983086983088983088
983085983089983088983086983088983088
983088983086983088983088
983089983088983086983088983088
983090983088983086983088983088
983091983088983086983088983088
983126983141 983154983161 983116 983151983159 983116 983151983159 983118983151 983149983145983150983137983148 983112983145983143983144 983126983141983154983161 983112983145983143983144 983109983160983156983154983137 983112983145983143983144
983108 983145 983155 983141 983139 983151 983150 983151 983149 983161 983151 983140 983123 983139 983137 983148 983141
983120983154983151983139983141983155983155 983117983137983156983157983154983145983156983161 983116 983141983158983141983148
983120983141983154983139983141983150983156983137983143983141 983107983144983137983150983143983141 983151983142 983108983145983155983141983139983151983150983151983149983161 983151983140 983155983139983137983148983141 983142983151983154 983123983156983137983150983140983137983154983140 983123983145983162983141 983120983154983151983146983141983139983156983155
983110983154983151983149 983123 983156983151 983113 983110983154983151983149 983123 983156983151 983117 983110983154983151983149 983123 983156983151 983116 983110983154983151983149 983123 983156983151 983126983116
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 89
[11] Diaz M and King J ldquoHow CMM ImpactsQuality Productivity Rework and the BottomLinerdquo STSC Crosstalk-the Journal of DefenseSoftware Engineering vol 15 no 3 pp 9-142003
[12] Donald H Krishnan M and Slaughter A
ldquoEffects of Process Maturity on Quality CycleTime and Effort in Software ProductDevelopmentrdquo Journal of Management Sciencevol 46 no 4 pp 451-466 2000
[13] El Emam K and Goldenson D ldquoAn EmpiricalReview of Software Process Assessmentsrdquo Advances in Computers vol 53 pp 319-4232000
[14] El Emam K ldquoTrialStat Corporation OnSchedule with High Quality and Cost Savings forthe Customerrdquo Journal of Software Technologyvol 10 no 1 pp 24-29 2007
[15]
Galin D and Avrahami M ldquoAre CMM ProgramInvestments Beneficial Analyzing Past StudiesrdquoSoftware IEEE vol 23 no 6 pp 81-87 2006
[16] Garmus D and Iwanicki S ldquoImprovedPerformance Should Be Expected from ProcessImprovementrdquo Journal of Software Technology vol 10 no 1 pp 14-17 2007
[17] Gibson D Goldenson D and Kost KldquoPerformance Results of CMMI-Based ProcessImprovementrdquo Technical Report SoftwareEngineering Institute CMUSEI-2006-TR-004ESC-TR-2006-004 2006
[18]
Girish H James J and Klein G ldquoSoftwareQuality and IS Project PerformanceImprovements from Software DevelopmentProcess Maturity and IS ImplementationStrategiesrdquo Journal of Systems and Softwarevol 80 no 4 pp 616-627 2007
[19] Goldenson D and Gibson D ldquoDemonstratingthe Impact and Benefits of CMMI An Updateand Preliminary Resultsrdquo Technical Report CMUSEI-2003-SR-009 Pittsburgh PASoftware Engineering Institute Carnegie MellonUniversity 2003
[20]
Goldenson D and Herbsleb J ldquoAfter theAppraisal A Systematic Survey of ProcessImprovement its Benefits and Factors thatInfluence Successrdquo Technical report CMUSEI-95-TR-009 Software Engineering InstituteCarnegie Mellon University Pittsburgh PA1995
[21] Herbsleb J and Goldenson D ldquoA SystematicSurvey of CMM Experience and Resultsrdquo in Proceedings of 18
th International Conference on
Software Engineering (ICSE) pp 323-330
Germany 1996[22] Herbsleb J Carleton A Rozum J Siegel J andZubrow D ldquoBenefits of CMM-Based SoftwareProcess Improvement Initial Resultsrdquo Technicalreport CMUSEI-94- TR-13 Software
Engineering Institute Carnegie MellonUniversity Pittsburgh 1994
[23] Humphrey S Snyder R and Willis RldquoSoftware Process Improvement at HughesAircraftrdquo IEEE Software vol 8 no 4 pp 11-231991
[24]
Jones L and Soule A ldquoSoftware ProcessImprovement and Product Line Practice CMMIand the Framework for Software Product LinePracticerdquo Technical Report CMUSEI-2002-TN-012 Pittsburg Pennsylvania CarnegieMellon University Software EngineeringInstitute 2002
[25] Liu A ldquoMotorola Software Grouprsquos ChinaCenter Value Added by CMMIrdquo Journal ofSoftware Technology vol 10 no 1 pp 18-232007
[26] Manish A and Kaushal C ldquoSoftware Effort
Quality and Cycle Time A Study of CMMLevel 5 Projectsrdquo IEEE Transactions on
Software Engineering vol 33 no 3 pp 145-1562007
[27] McGibbon T ldquoA Business Case for SoftwareProcess Improvementrdquo Final Report Contract Number F30602-92-C-0158 Data and AnalysisCenter for Software Kaman SciencesCorporation Utica New York 1996
[28] McGibbon T Grader M and Vienneau R ldquoTheDACS ROI Dashboard-Examining the Results ofCMMIreg Improvementrdquo Journal of Software
Technology vol 10 no 1 pp 30-35 2007[29] Memon G ldquoInfluence of Process Maturity Level
on Software Development Effort of StandardSize Projectsrdquo Journal of Information andCommunication Technology vol 2 no 1 pp 10-18 2008
[30] Paulk M ldquoHow ISO 9001 Compares with theCMMrdquo IEEE Software vol 12 no 1 pp 74-83 1995
[31] Paulk M Weber C Curtis B and Chrissis MThe Capability Maturity Model Guidelines for Improving the Software Process Addison-Wesley 1995
[32] Peter J and Rohde L ldquoPerformance Outcomesof CMMI-Based Process Improvementsrdquo Journal of Software Technology vol 10 no 1 pp 5-8 2007
[33] Pyzdek T The Six Sigma Handbook TheComplete Guide for Greenbelts Blackbelts and
Managers at All Levels McGraw-Hill 2003[34] Sapp M Stoddard R and Christian T ldquoCost
Schedule and Quality Improvements at WarnerRobins Air Logistics Centerrdquo Journal of
Software Technology vol 10 no 1 pp 10-132007[35] Staple M and Niazi M ldquoSystematic review
Systematic Review of OrganizationalMotivations for Adopting CMM-Based SPIrdquo
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences
8162019 Maged Yahya COCOMO Paper
httpslidepdfcomreaderfullmaged-yahya-cocomo-paper 99
Journal of Information and Software Technologyvol 50 no 7-8 pp 605-620 2008
[36] Wohlwend H and Rosenbaum SldquoSchlumbergerrsquos Software ImprovementProgramrdquo Journal of IEEE Transactions onSoftware Engineering vol 20 no 11 pp 833-
839 1994[37] Yi T Sheng T Ching C and Huang SldquoAssessing the Adoption Performance of CMMI-Based Software Process Improvement in 18Taiwanese Firmsrdquo Journal of Software Engineering Studies vol 1 no 2 pp 96-1042006
Majed Alyahya is a researcher inthe area of software engineering Heholds PhD degree in ComputerScience from University of Malaya
Malaysia in 2010 He received hisBachelor degree in ComputerScience from Zarqa Private
University Jordan in 2002 and Master degree inComputer Science from University of Jordan in 2004His research interests include project managementsoftware process models software process assessmentand improvement and software cost estimation
Rodina Ahmad is senior lecturer insoftware engineering and a researchmember of ICT cluster at the
University of Malaya Malaysia Sheholds PhD degree in informationsystem from the National Universityof Malaysia in 2006 She obtained
her bachelor degree and master in computer sciencefrom Rensselaer Polytechnic Institute United StatesDr Rodina has more than 19 years of experience inteaching and researching in the area of SoftwareEngineering and Information Systems She has published many conference and journal papers at thenational and international level At present DrRodina supervises seven PhD students in requirementsengineering education software release planningCMMI framework evaluation software dynamicmetrics and information system She also leads andteaches courses at both BSc and MSc levels incomputer science and software engineering
Sai Lee is a professor at Faculty ofComputer Science and InformationTechnology University of MalayaShe obtained her Master ofComputer Science from Universityof Malaya in 1990 her Diplocircme
drsquoEacutetudes Approfondies (D E A) inComputer Science from University of Pierre et MarieCurie (Paris VI) in 1991 and her PhD degree inComputer Science from University of Pantheacuteon-Sorbonne (Paris I) in 1994 Her current researchinterests include Software Reuse Application andPersistence Frameworks Requirements and DesignEngineering Object-Oriented Techniques and CASEtools She has published more than 80 research papersin local and international journals and conferences