Bug Report Analytics for Software Reliability Assessment using Hybrid Swarm -- Evolutionary Algorithm

Background: With the growing advances in the digital world, software development demands are increasing at an exponential rate. To ensure reliability of the software, high-performance tools for bug report analysis are needed. Aim: This paper proposes a new ‘Iterative Software Reliability’ model based on one of the most recent Software Development Life Cycle (SDLC) approach. Method: The proposed iterative failure rate model assumes that new functionality enhancement occurs in each iteration of software development and accordingly design modification is made at each stage of software development. In terms of defects, testing effort, and added functionality, these changing needs in each iteration are reflected in the proposed model using iterative factors. The proposed model has been tested on twelve Eclipse and six JDT software failure datasets. Proposed model parameters have been estimated using a hybrid swarm-evolutionary algorithm. Results: The proposed model has about 32% and 55% improved efficiency on Eclipse and JDT datasets respectively as compared to other models like Jelinski Moranda Model, Shick-Wolverton Model, Goel Okumotto Imperfect Model etc. Conclusion: In each analysis done, the proposed model is found to be reaching acceptable performance and could be applied on other software failure datasets for further validation.