Software Bug Prediction Prototype Using Bayesian Network Classifier: A Comprehensive Model

Abstract

Software bug prediction becomes the vital activity during software development and maintenance. Fault prediction model able to engaged to identify flawed software code by utilizing machine learning techniques. Naive Bayes classifier has often used times for this kind problems, because of its high predictive performance and comprehensiveness toward most of the predictive issues. Bayesian network(BN) able to construct the simple network of a complex problem using the fewer number of nodes and unexplored arcs. The dataset is an essential phase in bugs prediction, NASA/Eclipse free-ware are freely available for better results. ROC/AUC is a performance measure for classification of fault-prone or non-fault prone, H-measure is also useful while prediction technique, we will explore every parameter and valuable expects for experiment perspective. © 2018 The Authors. Published by Elsevier Ltd.