MULTIPARTITE BASED TEST CASE PRIORITIZATION USING FAILURE HISTORY

Abstract

Test case prioritization reorders test case execution sequence based on their early fault detection capability. In regression testing when new version is released, both current and previous version test cases are executed to cross check the desired functionality. Historical failure data ensures the previous fault detection information, which leads the potential faults in new version. Most of the existing prioritization strategies are developed either similar code or requirement coverage based, where some incorporated historical failure data. Clustering similar test cases may detect similar fault consecutively, which never ensure the maximum variety of fault coverage. This paper presents a prioritization approach based on multipartite graph clustering using historical data analysis to detect faults of different code coverage in minimum test case execution. In this strategy, test cases within each partite are distinct in terms of function coverage, which ensures dissimilarity among them. Proposed scheme is evaluated using well established Defects4j dataset, and it has reported that multipartite algorithm performs better than normal ordering, random, similarity clustering, and dissimilarity based test case prioritization approaches. The proposed method performs 42.85% and 14.28% better than similarity and dissimilarity based approaches respectively in terms of early fault detection.