A Single Change Point Hazard Rate Software Reliability Model with Imperfect Debugging

Abstract

Software reliability models with imperfect debugging can characterize the quality of software fault removal, in addition to the rate of faults discovered. Change points are therefore useful to quantity such changes in the fault discovery rate. To address these issues, this paper develops a hazard rate model based on the widely studied Jelinski-Moranda model, introducing two additional parameters, namely an imperfect debugging parameter and a single change point parameters. We compare the proposed model with simpler models to show these additional parameters are justifiable. Both information theoretic and predictive measures of goodness of fit are applied to demonstrate these additional parameters are appropriate on some data sets.