Estimating Change-Points Based on Defect Data

Software reliability is key in measuring software quality. Software Reliability Growth Models (SRGM), are widely used to predict growth in defect data. Changes due to major fixes or upgrades that cause change in the failure distribution, are called change-points. The number and locations of change-points affect reliability modeling. In our paper we examine the different change-point estimation methods used in software reliability modeling and compare them to the underlying change data that caused the change in defect distribution. The results show that data analysis of change in lines of code provides a credible indication of the existence of change-points.