Predicting software defects using machine learning techniques

A huge variety of software systems are relied upon in such domains as aviation, healthcare, manufacturing and robotics, and therefore, h systems and that they are reliable. Software defect prediction helps improve software reliability by identifying potential bugs during software maintenance. Traditionally, the focus of software defect prediction was on the design of static code metrics, which help with predicting the defect probabilities of a code when input into machine learning classifiers. While machine learning techniques such as Deep Learning technique, Ensembling, Data Mining, Clustering and Classification are known to help predict the location of defects in code bases, researchers have not yet agreed on which is the best predictor model. This paper will use 13 software defect datasets in evaluating the performance of the different predictor models. The results show that consistency in high accuracy prediction was achieved using Ensembling techniques. © 2020, World Academy of Research in Science and Engineering. All rights reserved.