Proposed Novel Hybrid Approach For Improvement in Bug Severity prediction

Abstract

In this research study an approach of creating dictionary of critical terms is proposed to assess the bug severity as severe and non severe. It is found that using different approaches of feature selection and classifier the pattern of accuracy and precision is approximately same. However Chi square test and KNN classifier give the maximum performance of precision and accuracy for the all four components. The proposed work will help Triager in classifying bugs based on severity and assigning these bugs to relevant developer.