Prediction of the Patch Release Date

Abstract

For a large and advancing Software System, the project group could get numerous bug reports over a long stretch of time. It is critical to accomplish a quantitative comprehension of bug fixing time. The capacity to predict bug fixing time can enable a project group better estimate programming support endeavours and better manage software programming ventures. In addition this time will be used to predict the release date of a minor version i.e. Patch. In industry when a client raises any bug, the undertaking supervisor needs to give them a date till which the minor variant will be released. The procedure for estimating the release date of the version needs to go through numerous stages like bug fixing time, smoke testing time lastly regression testing time. So we exhibit an effort that consequently predicts the fixing time. Our procedure uses existing issue following frameworks i.e. when a new bug report is generated the title and the description is extracted from the report that the bug with the similar title and description is searched from the database and here we have used Lucene framework for finding the bugs that have text similarity with the new bug report and used their time for prediction. In this approach, we have used Support vector Machine technique to query the database of resolved issues for textually similar reports. We also increase the reliability of our predictions by extending the SVM approach to explicitly state when there are no similar issues. Here this approach helps us for the early estimation of the bug fixing time, better assignment of the issues and predicting and scheduling the stable releases. Here we have assessed our approach utilizing the information from the RPAS (Retail Predictive Application Server) Project of an Oracle. Given an adequate number of issues reports, our programmed predictions are near the real exertion.