statistical quality techniques to service science and engineering

Prof. Fugee Tsung, HKUST 1

Statistical Quality Techniques to Service Science and Engineering

Fugee Tsung 宗福季 Professor and Head, Dept. of Industrial Engineering and Logistics Management

Director, Quality Lab, HKUST 香港科技大学工业工程与物流管理系教授主任质量实验室主任


Service Science and Engineering Driven by a new business environment including

globalized economy, business automation, and business and technology innovations, the service sector keeps growing and now accounts for more than 50 percent of the labor force in the developed economies. It reaches as high as 80 percent in the United States and the United Kingdom.

This unprecedented growth is changing the way companies organize themselves, resulting in a ripple effect in industries and universities with close ties to these organizations.

With the shift in economic focus from manufacturing to service, industrial and academic research facilities now need to apply more scientific rigor to the practices of service, such as discovering better methods to use statistics and mathematical optimization to increase quality, productivity, and efficiency to meet the challenges.


Quality Techniques and Six Sigma in Service In 1997, Citibank hired Motorola University to teach Six Sigma

defect reduction and cycle time reduction (CTR) to its employees. Most people think reducing cycle time applies only in the

manufacturing sector, but Citibank found CTR to be extremely useful in financial areas, such as consumer banking and emerging markets.

Citibank began its quality training initiative in 1997. From May 1997 to October 1997, more than 650 senior managers were trained. Between November 1997 and the end of 1998, another 7,500 employees attended sessions as part of senior-manager-led teams. By early 1999, 92,000 employees worldwide had been trained.

American Express, Bank of America, JPMorgan, Chase, Merrill Lynch and Vanguard are in various stages of Six Sigma deployment.

Statistical Quality Controlto Service Applications


Statistical Quality Control: Basic Idea

3020100

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5

Sample Number

Sam

ple

Mea

n

Mean=0

UCL=1.342

LCL=-1.342

?y

In most SQC applications it is assumed that the quality of a process or product can be adequately represented by the distribution of a univariate quality characteristic or by the general multivariate distribution of a vector consisting of several quality characteristics (Woodall et al. 2004).

),(~

),(~222

211

Ny

Ny

i

i

.,...,1

;,...,2,1

nifor

ifor

I. SQC/SPC for Profiles


Profile SPC: A Different Idea

3020100

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5

Sample Number

Sam

ple

Mea

n

Mean=0

UCL=1.342

LCL=-1.342

?yx

In many practical situations, however, the quality of a process or product is better characterized and summarized by a relationship between a response variable and one or more explanatory variables (Woodall et al. 2004).

)( ii xfy


Time in months

1.3

1.5

1.7

Jan 0 May 0 Sep 0 Jan 1 May 1 Sep 1

Flat

Time in months

01

23


Flat + Outlier

Time in months

0.5

1.0

1.5

2.0

2.5


Flat + Change Point

Time in months

60

65

70

75

80

85


Trend

Time in months

0.0

0.5

1.0

1.5

2.0

2.5


Trend + Outlier

Time in months

1.6

2.0

2.4

2.8


Trend + Change Point

Time in months

05

10

15

20

25


Growth

Time in months

78

91

011

12

13


Unknown

Time in months

23

45


Volatility

Profiles in service:Telecom Monthly Usage

Tsung, F., Zhou, Z.H., and Jiang, W., " Applying Manufacturing Batch Techniques to Fraud Detection with Incomplete Customer Information ," IIE Transactions, 2007.


Challenges of profile SPC to service applications Mostly transactional data. Many profile data in service are high-

dimensional and mixed-type. Time-varying consumer behavior in

individual profiles for fraud detection (human element/human impact; customer involvement)

…


Recent results in Profile SPC Tsung, F., Zhou, Z.H., and Jiang, W., " Applying Manufacturing

Batch Techniques to Fraud Detection with Incomplete Customer Information ," IIE Transactions, 39, 671-680, 2007.

Zou, C., Tsung, F., and Wang, Z., " Monitoring General Linear Profiles Using Multivariate EWMA schemes ," Technometrics, 49, 395-408, 2007.

Zou, C., Wang, Z., and Tsung, F., " A Self-Starting Control Chart for Linear Profiles ," Journal of Quality Technology, 39, 364-375, 2007.

Zou, C., Tsung, F., and Wang, Z., " Monitoring Profiles Based on Nonparametric Regression Methods ," Technometrics, 50, 512-526, 2009.

Tsung, F., Discussion on " Nonparametric Profile Monitoring By Mixed Effects Modeling by Qiu, Zou and Wang ," Technometrics, 52, 283-285, 2010.

II. SQC/SPC for Multistage Processes


A Multistage Process

3020100

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5

Sample Number

Sam

ple

Mea

n

Mean=0

UCL=1.342

LCL=-1.342

?ykyk-1

Stage 1 Stage NStage kStage k-1


An Example in Service – Hong Kong International Terminal (HIT): Vessel Discharging Process

The efficiency/quality of the whole process depends on the efficiency/quality of each operating stage.

P lacem en ton to a tru ck

T ran s p ortationfrom a b erth to a

yard

U n load from atru ck b y fork or

yard cran e

A llocation in th eres erv ed h all

C on tain erd is ch arg in g

b y a q u ay cran e

From HIT report.


Challenges of multistage SPC to service applications Most stages/business functions are

intangible. Unknown/complex stage-wise relationship. Measurement are not in place to provide

information at each stage. Network monitoring …


Recent results in Multistage SPC

Xiang, L. and Tsung, F., " Statistical Monitoring of Multistage Processes Based on Engineering Models ," IIE Transactions, 40, 957-970, 2008.

Zou, C., Tsung, F., and Liu, Y., " A Change Point Approach for Phase I Analysis in Multistage Processes ," Technometrics, 50, 344-356, 2008.

Zou, C. and Tsung, F., " Directional MEWMA Schemes for Multistage Process Monitoring and Diagnosis ," Journal of Quality Technology, 40, 407-427, 2008.

Li, Y. and Tsung, F., " False Discovery Rate-Adjusted Charting Schemes For Multistage Process Fault Diagnosis and Isolation ," Technometrics, 51, 186-205, 2009.

Jin, M. and Tsung, F., " A Chart Allocation Strategy for Multistage Processes ," IIE Transactions, 41, 790–803, 2009.

Zhu, K., Zhang, R. and Tsung, F., " Pushing Quality Improvement along Supply Chains ," Management Science, 53, 421-436, 2007.

Thank you!

Fugee TsungProfessor and Head, IELM, HKUSTFellow, Institute of Industrial Engineers (IIE) Fellow, American Society for Quality (ASQ) Fellow, the Hong Kong Institution of Engineers (HKIE) Six Sigma Master Black Belt, ASQQuality Lab -> http://qlab.ielm.ust.hk

statistical quality techniques to service science and engineering

Documents