Predicting Fault-prone Software Module Using Data Mining Technique and Fuzzy Logic

Abstract

This paper discusses a new model towards reliability and quality improvement of software systems by predicting fault-prone module before testing. Model utilizes the classification capability of data mining techniques and knowledge stored in software metrics to classify the software module as fault-prone or not fault-prone. A decision tree is constructed using ID3 algorithm for existing project data in order to gain information for the purpose of decision making whether a particular module id fault-prone or not. The gained information is converted into fuzzy rules and integrated with fuzzy inference system to predict fault-prone or not fault-prone software module for target data. The model is also able to predict fault-proneness degree of faulty module. The goal is to help software manager to concentrate their testing efforts to fault-prone modules in order to improve the reliability and quality of the software system. We used NASA projects data set from the PROMOSE repository to validate the predictive accuracy of the model.