Automatic Event Detection for Software Product Quality Monitoring

Collecting product metrics during development or maintenance of a software system is an increasingly common practice that provides insight and control over the evolution of a product's quality. An important challenge remains in interpreting the vast amount of data as it is being collected and in transforming it into actionable information. We present an approach for discovering significant events in the development process from the associated stream of product measurement data. At the heart of our approach lies the view of measurement data streams as functions for which derivatives can be calculated. In a manner inspired by Statistical Process Control, a certain number of data points are then selected as events worthy of further inspection. We apply our approach in an industrial setting, namely as support to the Software Monitoring service provided by the Software Improvement Group. In particular, we report on an evaluation of an alert service that continuously checks for events in over 400 monitored software systems.