Software Defect Prediction for Quality Improvement Using Hybrid Approach

ABSTRACT In today’s computing environment, software systems have become increasingly complex and versatile. Therefore it is necessary to continuously identify and correct software design defects. Software defect prediction plays an important role in improving software quality and it helps to reduce time and cost for software testing. The Software defect prediction is a method which predicts defects based on historical database. Different machine learning techniques are used to predict software defects from historical databases. The paper mainly focuses on generating accurate rules for software defect prediction System. For this purpose, K-means clustering technique is used for discretization. Then association rule mining is applied to generate rules in large volumes of data using Apriori algorithm. Software defect prediction system has been experimented on open sources NASA defect dataset. It contains software metric data and error data at the function or method level. In this paper comparison of proposed approach with existing approaches is discussed and the results show that proposed method is generated only interesting rules therefore it is effective for software defect prediction.