A novel framework for software test suite enhancement using data mining techniques

Abstract

Software testing is the process used to execute a program or framework with the intention of finding errors. It ordinarily incorporates three stages: the creation of test cases, execution of test cases and verification of actual outputs against the expected ones. A Test case is a set of conditions that contains test sequences with expected output and actual output. The execution of all the test cases requires lot of time and effort. Hence, the test case selection techniques aim to eliminate redundant or obsolete test data, by selecting and executing only a small subset of tests and it is essential in the reduction of time and effort spent for the testing process. Since many of the existing research works focus on software test suite reduction, Greedy Approaches and Evolutionary Computation methods have been employed to achieve optimization in the testing process. To the best of the gathered knowledge, none of the approaches have proposed the enhancement of the test suite by predicting the faulty/invalid test cases automatically as well as reduction in the size of the test suite, reduction of time in the testing process and elimination of redundant test cases. It has been the motivation factors for this research work. The contribution of this research work includes the development of a novel software test suite enhancement framework using data mining techniques such as Classification, Clustering and Association Rule Mining. In the first technique, Classification based test suite enhancement using Decision Tree Algorithm has been developed for classifying each test cases as valid and invalid test cases. It also predicts valid and invalid test cases automatically, when the testing process is executed against Software under Test (SUT). It will be helpful for the tester for testing the software with valid test cases without wasting testers time. The next technique, namely Clustering based test suite enhancement using k-means algorithm has been developed for newline