Effects of Class Imbalance in Test Suites: An Empirical Study of Spectrum-Based Fault Localization

Abstract

Spectrum-based fault localization techniques (SBFLs) focus on the utilization of execution information provided by large number of test cases. How to use the cases effectively is an important issue for SBFLs. Related studies mainly focus on the effects of test suites' sizes on localization efficiency, while ignoring the class composition, especially a class imbalance phenomenon existing in test suites. In this paper, an empirical study is undertaken to explore two questions: (1) Does the class proportion between passed and failed cases have effects on SBFLs? (2) If yes, what are the effects? The research may construct a bridge in connecting software fault localization and class imbalance learning.