Mining Objective Process Metrics from Repository Data

The configuration management repository includes abundant data not only on configuration items, but about the process itself. But meaningful information about the software process is often hidden. This paper presents a method of extracting software process metrics from software repositories. More specifically, the metrics presented use data from the bug or task tracker and from the configuration management event log. The metrics are presented in graphic forms common to traditional and lean project management practices. The metrics presented here are empirical – not subject to bias in reporting or interpretation, and focused on measuring the process itself - not the developers and artifacts. They are derived from data that commonly exist in project software repositories, and thus can be collected with little or no cost. The metrics are illustrated with real software development repository data collected from a large industry project over a time span of several years