A Hybrid Ensemble Method for Accurate Software Defect Prediction

The software defect prediction applies in improving the quality of the software delivered and the cost of testing the software as the testers will only test the modules which are defected. The Synthetic Minority Over-Sampling Technique (SMOTE) and Particle Swarm Optimization (PSO) are used in this classification unlike CM1 or JM1 because the researchers of such issues as the unequal representation of the classes or the most effective selection of features should be allowed. In case of a fused Stacking Classifier with Decision Tree, the Random Forest, and the Light GBM models, which were also rather effective and applied to all the datasets, respectively. The additional proposals are provided to render the proposed approach more accurate due to the assistance of the ensemble learning that presupposes the synthesis of the most effective aspects of multiple classifiers. The proposed model has also undergone the relevant tests and it is determined that it was a good and sound model. These results also indicate that it has a high degree of locating the compromised software modules as well as the software modules prediction. This execution is aimed to have an advantage to the software fault forecasting, in such a manner, that it is an inexpensive and swift approach of detecting errors at the initial stage. This will lead to the fact that the usage and quality of the testing resources will be improved.