Prediction of Defective Artifacts by Removing Redundant Metrics in Software Development Life Cycle (SDLC)

Software defect prediction (SDP) models aid software testing teams in forecasting faulty artifacts prior to actual testing. Software Development Process (SDP) models aim to comprehend previous shortcomings based on software metrics data that resulted in faulty modules. Previous research has determined that defect prediction models constructed without eliminating superfluous metrics yield unforeseen outcomes. This study examines the effect of superfluous metrics on the defect prediction performance of different software development processes. Through the analysis of five publicly available datasets containing system defects, it was determined that significant portions of the metrics examined were found to be superfluous and had only a negligible impact on predictive capabilities. The aforementioned observation leads to the inference that the development of a proficient defect prediction model necessitates the elimination of superfluous metrics from the datasets.