Predicting the Severity of Closed Source Bug Reports Using Ensemble Methods

摘要: Severity tells about how urgent given bug is to be fixed. There are large numbers of identified during software development and maintenance for each the report will submitted. Bug gives very important information such as Description, Severity, Priority, Date Time report, etc. Even though there clear guidelines present assign severity, inexperienced busy test engineer may make mistake in correctly identifying severity closed source development. Automatic prediction helps saving time resources. In this paper, dataset (PITS) taken NASA projects from PROMISE Repository. Predicting done using Bagging, Voting, Adaboost Random forest ensemble methods. The result shows bagging better accuracy than other algorithms. Two preprocessing techniques, i.e., Information gain Chi-square considered data reduction. slightly good accuracies over chi-square.