OPTIMAL RULE SELECTION BASED DEFECT CLASSIFICATION SYSTEM USING NAVE BAYES CLASSIFIER

摘要

Defect Management process plays key role during Software Testing life cycle, since one of the objectives of testing is to find defects, the discrepancies between actual and expected outcomes need to be logged as defects or bugs or incidents. In order to manage all defects to completion, an organization should establish a process and rules for classification. Software defects are more expensive and time consuming. The cost of finding and correcting defects represents one of the most expensive software development activities. In our previous work, the defect classification was done by association rule mining and decision tree algorithm. Association rule mining algorithm sometimes leads to insignificant rules. So it is very difficult to classify the defects based on these insignificant rules. In order to avoid such issues, we have to optimize the rules before classification based on support and confidence value. In the present work, the rules were extracted from the database using association rule mining. The association rules are optimized using ABC algorithm. Then the defects were classified using Nave bayes classifier. This performs defect classification in an efficient way. Finally the quality will be assured by using various quality metrics such as defect density, Sensitivity etc.