Application of Adaptive Neural Fuzzy Inference System for the Prediction of Software Defects

One of the major challenges in the process of software development is the prediction of software defects for the reduction of cost for the implementation of any software. Great cut off cost is done in software field because of the implementation of the predicting defective modules. Many scholars have used many techniques from various aspects such as data mining for the prediction datasets of the software defects that can be downloaded from NASA repositories. The datasets downloaded are pretty much imbalances on their own. In the current study we compute the Adaptive inference for neural fuzzy systems. Subtractive Clustering Method (SCM) has given the base structure for the initial fuzzy inference system and innovative rule for learning is used for their training. The balance in the datasets is lost but the performance can be understood with the help of the values (AuC) for the classifier. We did the computation and compared neural networks that are cost sensitive with the FIS. The section which consists of the results contains the curves of operating characteristics of the receiver which are computed and presented. The cost sensitive methods compared with the FIS, and the drawn results are found to be satisfactory.