Test case prioritization to examine software for fault detection using PCA extraction and K-means clustering with ranking

Many software-related failures or faults were caused as the consequence of not detecting it early and prevailing constraints of time and supplies available during any software examination. Having identified the challenges in the regression software testing of any software, many have started moving their attention towards the test cases or else validation suites prioritization. In this work, we have taken the Firefox bugs Report as the input Dataset comprising 12,486 test cases in it. Dealing with these kind of greater number test cases might consume our times and resources to the core for which the work is proposed. The dataset will be subjected to pre-processing operation to remove the unwanted contents in the bug reports like the assertions in some cases. Necessary features are then selected by using the algorithm called PCA, and attributes for clustering and prioritization will be determined by using the Dimensionality reduction. After the feature selection process, we make use of the agglomerative K-means clustering algorithm which helps to form clustered groups. After clustering process, we apply the ranking algorithm and prioritized the test cases in the clusters by using it. Finally, the performance of this work was analyzed and cross-verified by analyzing the priority ranking for clusters, sum of the distinct faults detected based on the number of test cases, adjacency matrix between test cases and fault detection rate.

Automated summary

This research utilizes Firefox bugs Report as the input Dataset and proposes a method for prioritizing test cases through preprocessing, feature selection, clustering, and prioritization steps.