Statistical Process Control for Software Inspections

article

By

Don Porter

|

April 24, 2001

Summary

This paper provides innovative guidelines for inspections by employing statistical process control (SPC).

The innovations include:

Tracking fault density = faults/(amount inspected) rather than the number of faults. Since the number of faults follows a Poisson probability law, the fault density obeys a related exponential distribution.
Tracking preparation and inspection efficiencies rather than rates. These are defined as (preparation time)/(amount inspected) and (inspection time)/(amount inspected). These variables fit gamma or lognormal distributions quite well.

Download Article:

XDD2469filelistfilename1_0.PDF

Topics:

process metrics reporting reviews testing

About The Author

Don Porter

Don Porter graduated from Northern Illinois University, where he received the Master of Arts Degree in Economics in 1978 and the Master of Science in Statistics in 1980. His undergraduate work focused on Economics and Mathematics. Don taught courses in Economics, Mathematics, and Statistics until 1985, when he entered the business world as a statistical consultant. He has worked for several Fortune 500 companies, including Continental Telephone, Ameritech, Abbott Labs, BP-Amoco and Motorola, his current employer. His clientele also included several smaller companies in the Chicago area. His experience includes statistical modeling, experimental design, sampling design, statistical quality control and process improvement. In his free time Don is a diligent student of the Bible as well as Ancient Near Eastern religions, history and culture. He likes to talk to people about these topics. He also keeps his pocket billiard skills honed on his basement pool table.