Hybrid Machine Learning Model for Software Defect Prediction

Abstract

A portion of software created to fulfill a particular purpose is known as software application. At the same time, engineering is focused on creating goods with specific technical methods and principles. Software defects can be anticipated at several stages, including data input and pre- processing, attribute extrication, and classification. This research study implements multiple classifiers to forecast software defects. This work makes use of several classifiers namely RF (random forest), GNB, Bernoulli NB, and MLP to forecast software faults. The development of an ensemble classifier increases the software fault's reliability. Class balancing and the Principal Component Analysis (PCA) approaches have been combined in the ensemble classifier that is being presented. Python is used for implementing the architecture that has been introduced. Diverse measures are used to examine the findings with regard to universal metrics (i.e. accuracy, precision, and recall).