Software Defect Prediction System using Multilayer Perceptron Neural Network with Data Mining

Abstract

Fault prediction in software systems is crucial for a ny software organization to produce quality and reliabl e software. Faults (defects) or fault-proneness of software modul es are to be predicted in the early stages of software life cycle , so that more testing efforts can be put on faulty modules. Vari ous metrics in software like Cyclomatic complexity, Lines of Code have been calculated and effectively used for predicting faul ts. Techniques like statistical methods, data mining, machine lear ning, and mixed algorithms, which were based on software metrics associated with the software, have also been used to predict software defects. Many works have been carried out i n the prediction of faults and fault-proneness of software systems using varied techniques. In this paper, an enhanced Multilayer Perceptron Neural Network based machine learning technique is explored and a comparative analysis is performed for the modeling of fault-proneness prediction in software s ystems. The data set of software metrics used for this research is acquired from NASA's Metrics Data Program (MDP).