A novel under sampling strategy for efficient software defect analysis of skewed distributed data

Abstract

The software quality development process is a continuous process which starts by identifying a reliable fault detection technique. The implementation of the effective fault detection technique depends on the properties of the dataset in terms of domain information, characteristics of input data, complexity, etc. The early detection of defective modules provide more time for the developers to allocate resources effectively to deliver the quality software in time. The class imbalance nature of the software defect datasets indicates that the existing techniques are unsuccessful for identifying all the defective modules. Misclassification of the defective modules in the software engineering datasets invites unexpected loses to the software developers. To classify the class imbalance software datasets in an efficient way, we have proposed a novel approach called as under sampling strategy. This proposed approach uses under sampling strategy to reduce the less prominent instances from majority subset. The experimental results confirm that the proposed approach can deliver more accuracy in predicting the modules which are error prone with less and simple rules.