Software Defect Data and Predictability for Testing Schedules

Software defect data are typically used in reliability modeling to predict the remaining number of defects in order to assess software quality and release decisions. However, in practice such decisions are often constrained by availability of resources. As software gets more complex, testing and fixing defects become difficult to schedule. This paper attempts to predict an estimated time for fixing software defects found during testing processes. We present an empirical approach that employs well-established data mining algorithms to construct predictive models from historical defect data. We evaluate the approach using a dataset of defect reports obtained from testing of a release of a large medical system. The accuracy obtained from our predictive models is as high as 93% despite the fact that not all relevant information was collected. The paper discusses detailed methods of experiments, results and their interpretations.