Selection of optimal software reliability growth model using a diversity index

Abstract

Software reliability growth models (SRGMs) have been arisen to estimate various criteria such as the number of errors remaining in the software, the software failure rate, and determining the reliability of the software. In general, SRGMs are dataset dependent and hence the selection of an optimal model for use in a particular application is considered an important issue in software reliability engineering. This study provides two multi-criteria decision-making methods for comparison and selecting the optimal SRGM for a particular dataset. The methods compute a weight, in terms of the degree of diversity, for each considered statistical criterion and provide a score for each SRGM, regarding the values of weights. Simplicity, weighting to criteria, and combining various descriptive and predictive aspects of a model in the process of the model selection are of the advantages of the methods.