An Experimental Study Using Machine Learning Techniques for Software Quality Prediction

Abstract

Software quality assessment is a necessary task at different phases of software development. It can be applied to project quality assurance practice design and benchmarking. Software quality was assessed using two methodologies in earlier research: multiple criteria linear programming and multiple criteria quadratic programming. The quality of C5.0, SVM, and neural mesh was also examined. These studies' accuracy is quite low. We attempted to increase the estimation accuracy in this work by utilizing pertinent information from a sizable dataset. To increase accuracy, we employed a correlation matrix and a non-selective approach. We also evaluated several new techniques that have worked well for previous forecasting assignments. Machine learning techniques including MLP Classifier, Random Forest, Decision Tree, XG Boost, and Logistic Regression and the data is subjected to Naive Bayes. forecast the quality of software and show how development parameters relate to quality. Results from the experiment indicate that machine learning algorithms are capable of accurately assessing the software quality level.