Naive Bayes Classification for Software Defect Prediction

Abstract

Software defects are an inevitable aspect of software development, exerting substantial influence on the reliability and performance of software applications. This research addresses the imperative need to enhance the prediction and monitoring of software defects within the software development domain. With a focus on system stability and the prevention of software malfunctions, this study underscores the significance of proactive measures, including robust software testing, routine maintenance, and continuous system monitoring. The central challenge addressed in this research pertains to the insufficient efficiency of predicting software defects during the development phase. To address this challenge, the study employs the Naive Bayes classification method. Test results conducted on the complete dataset reveal that the Naive Bayes method yields classifications with an exceptionally high accuracy rate, reaching 98%. These findings suggest that the method holds great potential as an effective tool for predicting and preventing software defects throughout the software development process. Additionally, through linear regression analysis, the model exhibits an intercept value of -0.09359968 and a coef coefficient of 0.00761893. The outcomes of this research bear significant implications for the implementation of the Naive Bayes method in software bug prediction analysis, particularly in the utilization of the Python programming language with the assistance of Google Colab. The adoption of this method can play a pivotal role in mitigating risks and elevating the overall quality of software during the developmental stages.