Authors : Satya Srinivas Maddipati, A. Yesubabu and G. Pradeepini
Abstract: The quality of software is improved by identifying defective prone modules which is influenced by various characteristics of software module like lines of code, Halstead metrics and cyclometric complexity values. There are various prediction models for identifying defective prone modules from these characteristics. In this study we are comparing neural networks and Adaptive Neuro Fuzzy Inference System (ANFIS) for software defect prediction. We applied gradient descendent learning for Neural Networks and Hybrid Leaning for ANFIS. The performance of the models are evaluated by using the metric Area under ROC curve (AuC) values. In these experiments, we considered Software Defect Prediction Datasets download from NASA repositories. The results of ANFIS are found satisfactory compared to Neural Networks.
Satya Srinivas Maddipati, A. Yesubabu and G. Pradeepini, 2017. A Comparative Analysis of Hybrid Learning over Back Propagation for Identifying Defective Prone Modules. International Journal of Soft Computing, 12: 199-203.