Clustering Driven Machine Learning Framework for Scalable Test Case Minimization and Optimization

Abstract

Introduction: Software testing is responsible for ensuring the quality and reliability of software, but long processing time and high resource consumption normally decelerate its efficiency. The paper is a comparative study of three research papers on test case reduction and optimization techniques for enhanced regression testing in agile software development environments. As an enhancement of regression test efficiency, researches address clustering techniques, which are K-means, FSK-means (Fractional Sigmoid K-means) and DBSCAN. The purpose of clustering techniques is to reduce the number of test cases that are redundant by keeping the high fault detection ratios. Among such issues in studies, issues of making accurate parameters for the clustering techniques require, constraint in the use of optimization techniques in manual testing, and computational complexity of metaheuristic techniques. Finally, the reduction of test cases is important and useful in the software testing, especially in the agile and in the industrial environment. Industrial optimization techniques, and metaheuristic techniques can provide a great improvement in the test execution efficiency and fault detection rates by using clustering techniques. Therefore, they provide means of performing regression testing with improved efficiencies and scalability, the consequences of which are better software quality and reliability.