Application and Benchmarking of Artificial Immune Systems to Classify Fault-Prone Modules for Software Development Projects

Çatal Ç., DİRİ B.

International Conference Applied Computing, Salamanca, Spain, 18 February 2007, pp.1-5, (Full Text)

  • Publication Type: Conference Paper / Full Text
  • City: Salamanca
  • Country: Spain
  • Page Numbers: pp.1-5
  • Yıldız Technical University Affiliated: Yes

Abstract

ABSTRACT Software quality assurance is a crucial activity to enhance and ensure the quality of software development projects. Necessary budget, personnel and resource should be allocated for quality assurance activities to minimize the operational level faults by identifying faulty modules. In this study, Artificial Immune Systems (AIS) have been investigated to classify fault-prone modules. AIS algorithms such as AIRS1, AIRS2, AIRS2 Parallel, Immunos1, Immunos2, Immunos99, CSCA, and CLONALG have been applied and investigated by varying user-defined parameters to reach the best classification accuracy for KC2 dataset which is a part of NASA Metric Data Program. Furthermore, ROC (Receiver Operating Characteristic) analysis has been used in this study.

KEYWORDS Fault-prone module classification, artificial immune systems, AIRS, Immunos, clonal selection, metrics.