Analyzing Fuzzy Estimation of Risks in a Scrum Team of a Global Software Project

Scaling Agile frameworks like SAFe, DAD or LESS help to eliminate or mitigate risks in global software development (GSD) projects. However, not all types of risks can be handled when using these frameworks. Additionally, the mitigation practices have long-term or mid-term effects. This means that even when using the aforementioned frameworks there are still many risks related to GSD projects that can affect mid-term or short-term project objectives. This paper presents a fuzzy algorithm based on linguistic input to estimate short-term (Sprint) risks for Scrum-Based GSD projects. This approach realizes the idea of explicit risk management applied to Scrum. The objective of the paper is to analyze the presented algorithm, where some steps are configurable, and select the best configuration. Different configurations of the algorithm are analyzed using post-factum data collected in a Scrum-based GSD. Algorithm quality is being assessed for these configurations and the best configuration is selected. The algorithm that includes both self-credibility and group-credibility assessment seems to be a promising solution for the given project data. Quantitative risk management applied for Sprint allows to adapt Sprint plan to the materialized risks, control short-term project objectives and increases chances to realize Sprint commitments. The suggested fuzzy-based algorithm fits well the Scrum-context. The results presented in this paper show that some elements of the algorithm are crucial for providing reasonable risk consequences assessments. Even when the inputs of the algorithm are imprecise and based on linguistic expressions, a quality increase compared to standard approaches is expected.