In the last two years, software bug prediction techniques have become increasingly accurate and it is now realistic to begin using them in software engineering practice. Bug prediction technology is introduced into two industrial software projects at separate companies. The first case study will give developers bug prediction information as a form of risk awareness, then explore how the developers react to this information. The second case study will predict the most bug prone files. Quality improvement resources will then be assigned to those files to remove bugs using a combination of inspections, testing, static analysis, and re-factoring. This project will provide increased understanding of how developers alter their development behavior when faced with high quality bug prediction information. It will also increase understanding of how the application of quality improvement resources on bug-prone files affects their quality in the near and long term. It will also yield improved understanding of how mixtures of quality improving techniques (inspections, testing, static analysis, re-factoring) combine to impact software quality. The project will also create improved bug prediction techniques that connect bug predictions with operational reliability (predicting high severity bugs) and provide developer-customized bug predictions.
