Hybrid Labels Are the New Measure!

Developing minimum viable products (MVPs) is critical for start-up companies to hit the market fast with an accepted level of performance. The US Food and Drug Administration mandates additional nonfunctional requirements in healthcare systems, meaning that the MVP should provide the best availability, privacy, and security. This critical demand is motivating companies to further rely on analytics to optimize the development process. In a collaborative project with Brightsquid, the authors provided a decision-support system based on analogical reasoning to assist in effort estimation, scoping, and assignment of change requests. This experience report proposes a new metric, change request labels, for better prediction. Using different methods for textual-similarity analysis, the authors found that the combination of machine-learning techniques with experts’ manually added labels has the highest prediction accuracy. Better prediction of change impacts allows a company to optimize its resources and provide proper timing of releases to target MVPs. This article is part of a special issue on Actionable Analytics for Software Engineering.