Standard Article

Software Module Risk Analysis

  1. Taghi M. Khoshgoftaar1,
  2. Naeem Seliya2

Published Online: 14 DEC 2007

DOI: 10.1002/9780470050118.ecse389

Wiley Encyclopedia of Computer Science and Engineering

Wiley Encyclopedia of Computer Science and Engineering

How to Cite

Khoshgoftaar, T. M. and Seliya, N. 2007. Software Module Risk Analysis. Wiley Encyclopedia of Computer Science and Engineering. .

Author Information

  1. 1

    Florida Atlantic University, Boca Raton, Florida

  2. 2

    University of Michigan—Dearborn, Dearborn, Michigan

Publication History

  1. Published Online: 14 DEC 2007

Abstract

Estimating the quality of software modules provides a practical way to assess their potential risk. A software quality estimation can be used to provide a targeted software quality improvement initiative. The aim is to provide the management team with a practical quality assessment of software modules for a cost-effective utilization of software quality improvement resources. In our previous and ongoing research, we have empirically investigated three different kinds of software quality models that are built using software measurement data—software quality classification, software fault prediction, and module-order modeling. The classification model predicts the class membership of software modules into the fault-prone and not fault-prone classes. The fault prediction model estimates the number of faults expected in the modules. A module-order model predicts the ranking of the modules with respect to their relative software quality. The preference among the three software quality models depends on the software quality improvement goals of the given software project team. We present the useful principles and a detailed overview of the three software quality estimation models, using a case study of a large-scale telecommunications software system. The classification model is built using the logistic regression technique, whereas the fault prediction model is built using the multiple linear regression technique. The module-order model is built using the estimations of the fault prediction model for ranking the software modules. We summarize our work with suggestions for software tools that can be used to build software quality estimation models.

Keywords:

  • software quality;
  • software measurements;
  • software faults;
  • software quality estimation;
  • software quality improvement