Decision Support System for Optimal Selection of Software Reliability Growth Models Using a Hybrid Approach

Abstract

A hybrid approach, namely “entropy-combinative distance-Based assessment (CODAS-E),” is proposed and presented to select and rank software reliability growth models based on multiple performance indexes, which is hitherto not applied in the open literature for the purpose. In the proposed hybrid approach, i.e., CODAS-E, the Shannon entropy approach is used to obtain the performance indexes’ priority weights and the CODAS method is used for optimum selection and ranking. The methodology is illustrated through two previously published different failure datasets. The ranking results depict that “Zhang-Teng-Pham” is the least suited model for software reliability estimation, whereas “Musa Okumoto” and “Yamada Imperfect debugging2” are best suitable for dataset-1 and dataset-2, respectively. The CODAS-E method is validated comparing with existing multicriteria decision-making methods; namely, technique for order preference by similarity to ideal solution and analytic hierarchy process. The significant contributions of the present research article include implementation of efficient, user-friendly, and time effective CODAS-E methodology to find the optimal model and the best overall ranking of employed models for any given dataset, and importance to the taxonomy of NHPP SRGMs rather than adding any new model. The presented model selection strategy will undoubtedly lead to high-quality software development.