An Assessment of Eclipse Bugs' Priority and Severity Prediction Using Machine Learning

The reliability and quality of software programs remains to be an important and challenging aspect of software design. Software developers and system operators spend huge time on assessing and overcoming expected and unexpected errors that might affect the users’ experience negatively. One of the major concerns in developing software problems is the bug reports, which contains the severity and priority of these defects. For a long time, this task was performed manually with huge effort and time consumptions by system operators. Therefore, in this paper, we present a novel automatic assessment tool using Machine Learning algorithms, for assessing bugs’ reports based on several features such as hardware, product, assignee, OS, component, target milestone, votes, and versions. The aim is to build a tool that automatically classifies software bugs according to the severity and priority of the bugs and makes predictions based on the most representative features and bug report text. To perform this task, we used the Multi-Nominal Naive Bayes, Random Forests Classifier, Bagging, Ada Boosting, SVC, KNN, and Linear SVM Classifiers and Natural Language Processing techniques to analyze the Eclipse dataset. The approach shows promising results for software bugs’ detection and prediction.