Supporting Software Decision Meetings: Heatmaps for Visualising Test and Code Measurements

To achieve software quality it is critical to quickly understand the current test status, its changes over time as well as its relation to source code changes. However, even if this information is available in test logs and code repositories it is seldomly put to good use in supporting decision processes in software development. The amount of information is often large, is time consuming to extract and hard to monitor. This case study shows how visualisation and correlation between software measurements can support improvement discussions. In particular, simple heat maps were found to be effective to visualize and monitor changes and identify recurring patterns in the development of a space-bourn, embedded control system. Statistical analysis quantified the correlation between different sources of development data and heat maps then effectively focused the attention of stakeholders to importants parts of the system. Here the visual analysis was focused on post-project, historical data but we discuss how early identification based on dynamic data analysis could support more effective analysis, planning and execution of quality assurance. Based on our findings we state requirements on such an online, visual analysis system and present a prototype implementation that can help software measurements better support value-based decisions in software development.