Designing a Multi-Class Classification Optimized Model for Requirement based Test Case Prioritization

Most of the times we have to test out the entire application functionality, for any code modification done to cater the need of the larger audiences or any bug fixes. This results to consumption of time and the effort to retest the application by executing all of the test suites. In such cases more often regression testing comes to the rescue, where prioritization techniques are being used to overcome the limitations of regression testing. Test Case Prioritization (TCP) usually means categorically ranking some test cases higher than others. The main goal of the TCP is to find the fault early in the testing process by scheduling the test cases with the help of Requirement based Test Case Prioritization Technique (RTCP) in order to increase effectiveness of regression testing. In this research study, we have developed a supervised machine learning based RTCP mechanism where the significant business requirement based relevant feature are considered for scheduling test cases according to their priority label as high, medium and low multi classes. The proposed model is validated with the two datasets collected from internet sources. The machine learning classifier k-Nearest Neighbor (K-NN), Decision Tree (DT), Random Forest (RF) Bagging and Boosting algorithm are utilized to evaluate the features for the test case prioritization. To enhance the performance of the model some hyper parameter settings are altered and found a drastic change in the results. To achieve the low cost and high fault detection rate in RTCP, an optimized model is designed which is validated by datasets and altered parameter settings. The experimental result demonstrates that RF classifier achieved the best performance among the other classifiers for predicting the prioritized test cases early to reduce the cost and time required for regression testing.