Software Process Simulation Based on Mining Software Repositories

Abstract

Software development processes and their quality depend on many evolutionary factors, such as project size and collaboration of the development team, and development strategies. For project managers it would be valuable to test the interplay of different possible future scenarios in advance, before making decisions about the further process. Our aim is to build such a tool based on agent-based simulation. In this paper, we present our simulation model and our approach in mining software repositories as input for it. The simulation model focuses on system growth, bugs lifetime and developer activity. The detection of patterns related to these topics allows us to estimate adequate simulation parameters close to reality. We evaluate this approach by comparing empirical and simulated data.