Hybrid software reliability prediction model based on residual errors

Abstract

Software reliability is one of the key features of software quality assurance. It is primarily associated with the defects of software, which indicates a major factor in software development. Software has become an indispensable investment for an organization. The users are more concerned about accurate and efficient failure prediction model. Early fault prediction model can reduce the cost of the test, increase reliability and improve the quality of software. Precise prediction is desired from the every model but no model has proved to be successful at efficiently and accurately developing software reliability model. In this paper, the authors have combined two most popular reliability prediction models and get the most accurate output from them by using Markov method based on their residual errors. Furthermore, experiment analysis and comparison carried out on U.S. Navy Tactical Data System, which shows the accuracy of the hybrid approach.