Fault Prediction Modelling in Open Source Software Under Imperfect Debugging and Change-Point

Abstract

Instant demand of products and services by technologically active users has increased the demand for open source software (OSS)-based applications. Unfortunately, with the complexity and lack of understanding of OSS-based systems, it becomes difficult for a testing team to remove the faults and the fault removal rate becomes low in comparison to what it should be. This also results in generating new faults during removal. Also, the rate at which the testing team detects/corrects fault need not be same during the entire process of testing due to various reasons viz. change in testing strategy, understanding of code, change in resources, etc. In the existing literature on OSS, authors have developed many models considering the above aspects separately. In this article, all of the above aspects have been combined to develop a general framework for predicting the number of faults in OSS. The comparison of eight models on the basis of their prediction capability on two well-known Open Source Software datasets is created and then ranked using normalized criteria distance approach.