WITHDRAWN: Machine learning approach for classification from imbalanced software defect data using PCA & CSANFIS

Abstract

Identifying Defective prone modules in a software is one of the important smart topics in the field of Software Engineering and Soft Computing. Many researchers applied various machine learning algorithms on identifying defective prone modules in software. In this proposed system, we are applying a Machine learning approach for Prediction of Software Defects to overcome from class imbalance problem. We applied principle component analysis as an attribute selection algorithm to identify most relevant attributes in identifying defective prone modules. Later, the data to be classified is imbalanced in nature. To overcome from imbalanced learning we are proposing Cost Sensitive ANFIS to derive a Sugeno Fuzzy Inference scheme for forecasting of Software defects. The performance of derived Fuzzy implication scheme was compared with the classifier used in literature survey. We observed Area under ROC curve has been improved by 5% with Sugeno Fuzzy Inference classier compared to the classifiers Neural Networks and Support Vector machines etc.