Software Defect Prediction Based on Classication Rule Mining

Abstract

There has been rapid growth of software development. Due to various causes, the software comes with many defects. In Software development process, testing of software is the main phase which reduces the defects of the software. If a developer or a tester can predict the software defects properly then, it reduces the cost, time and eort. In this paper, we show a comparative analysis of software defect prediction based on classifcation rule mining. We propose a scheme for this process and we choose different classication algorithms. Showing the comparison of predictions in software defects analysis. This evaluation analyzes the prediction performance of competing learning schemes for given historical data sets(NASA MDP Data Set). The result of this scheme evaluation shows that we have to choose different classifer rule for different data set.