Performance of Software Quality Prediction with Machinelearning Methods

The prediction of software quality through machine learning (ML) is an expanding area that focuses on applying different ML algorithms to anticipate the quality of software systems. Software quality assessment is a crucial task required at different phases of software development. It can be utilized for organizing the quality assurance practices of the project and for comparison purposes. In prior studies, two approaches (Multiple Criteria Linear Programming and Multiple Criteria Quadratic Programming) were employed to assess software quality. Additionally, C5.0, SVM, and neural networks were tested for quality assessment. These studies exhibit fairly low accuracy levels. In this research, we sought to enhance estimation precision by utilizing significant attributes from a substantial dataset. We employed a feature selection approach and a correlation matrix to achieve improved accuracies. Moreover, we have tested new techniques that have proven effective for different prediction challenges. Machine learning techniques like XGBoost (Gradient Boosting), Random Forest, Decision Tree, Logistic Regression, Bagging Classifier, and K Nearest Neighbour are utilized on the data to forecast software quality and uncover the relationship between quality and development characteristics. The experimental findings indicate that machine learning algorithms can accurately predict the quality level of software.