AN OPTIMIZED MUTATION TESTING USING HYBRID METAHEURISTIC TECHNIQUE WITH MACHINE LEARNING FOR SOFTWARE DEFECT PREDICTION

Abstract

Software defect prediction model based on the mutation testing is a pioneering method for the fault-based unit testing in which faults are detected by executing certain test data. This paper presents an Optimized Mutation Testing (OMT) technique based software defect prediction model using the concept of hybrid metaheuristic technique. Here, hybridization of OMT with Enhanced Learning-to-Rank (ELTR) is used for the feature extraction from mutation testing based data generation mechanism. In the proposed approach, first hybrid technique is used for the test data feature extraction then this data is exercised to cover all mutants present in the specific program under test and then machine learning based Random Forest as an ensemble classifier is used as a classifier. The proposed method can improve the testing as well defect prediction efficiency by deleting the redundant test data. In this research work, two models are implemented for the software defect prediction using the ELTR and LTR. At last, the performance parameters such as Detection Rate, Defect Predication Value, Execution Time, Percentage of Fault Negative Rate and Percentage of Fault Rate are measured and compared with the existing work to validate the proposed model.