Software Hazard Rate Modeling with Multiple Change-point Occurrences

We propose a software hazard rate modeling framework with the effect of multiple change-point occurrences for developing more plausible software reliability growth models reflecting actual environment in the testing-phase of a software development process. Especially, our modeling approach enables us to develop a software hazard rate model with the effect of multiple change-point occurrences by assuming the hazard rate before the first change-point occurrence. And we check that our multiple change-point model has better fitting and predictive performance than the corresponding non and single change-point models by conducting goodness-of-fit comparisons applying using actual data.