Piecewise Congruence Regressed Indexive Extreme Learning Classifier for Software Fault Prediction

Software fault prediction is a significant task in software development to discover faults early. It is the process of developing models that can be used by the software practitioners in the early phases of software development life cycle for detecting faulty constructs such as modules or classes. Therefore, early fault prediction is a critical and challenging task faced by the Project managers. Several kinds of approaches were utilized to predict software faults. In this work we propose a Piecewise Congruence Regressed Indexive Extreme Learning Classifier (PRILEC) to predict the software faults accurately. The process consists of two stages namely feature selection or software metric selection and classification. In feature selection process, congruence correlative piecewise regression method is utilized to extract the most relevant features from the given input dataset. In the next phase, statistical indexive levenberg extreme learning classifier is utilized to classify the fault prediction with better accuracy. The testing and training data analysis in extreme learning classifiers is evaluated using Camargo’s statistical index. Hardlimit activation function is utilized to identify the faulty or non-faulty software code. The least square problem can be minimized using Levenberg-Marquardt algorithm and this algorithm can obtain the better classification results. The performance of the proposed approach is evaluated using software fault prediction data analysis dataset. The evaluation metrics such as precision, recall, F-measure, and specificity were used to assess the performance of the proposed algorithm. It was observed that compared with the state-of-the-art traditional methods (Linear regression), proposed technique increases data accuracy of software fault prediction. The system reduces the fault detection time by 4%, 2%, 2%, 2%, 29% and 21% respectively.