A Novel Framework to Evaluate Software Reliability Prediction Models Using Multi-Criteria Decision-Making

Software Reliability is one of the most important factors to consider when assessing a software product's quality. Numerous researchers have developed several software reliability prediction (SRP) models to aid in maintenance and replacement. However, each model may have varying capacity to predict software reliability in the context of many competing accuracy criteria. As the evaluation of SRP models involves various criteria, the problem of selecting the best SRP model can be modelled as the multi-criteria decision making (MCDM) problem. This study's goal is to propose a framework based on MCDM to assess the effectiveness of several SRP models taking into account a variety of competing accuracy measures as a whole. An experimental study was carried out for assessing the performance of ten SRP models over a software failure dataset considering four performance measures altogether in order to validate the proposed approach. Based on MCDM ranking, the model SOMFTS is suggested as the most suitable SRP model. According to the study's experimental findings the proposed MCDM based method may be successfully utilized as a decision-making tool to choose the best prediction model for predicting the software reliability.