A Recommender System-Based Approach to Risk Management in Scrum Projects

Abstract

Risk management is essential in software project management. It includes activities such as identifying, measuring, and monitoring risks. The increasingly popular agile methods don’t offer specific activities to manage risk. The lack of risk management or its inadequate application is one of the reasons for the failure of software development projects. Therefore, we developed an approach to risk management in software development projects that use Scrum. The proposed approach provides a set of risk management practices and an iterative life cycle. Along with this approach, we developed a recommendation algorithm to assist decision-making when identifying risks. Thus, we performed an offline evaluation to verify the best configuration for the recommendation algorithm that will accompany our approach. We chose Manhattan similarity based on the experimental results collected, with a precision of 45%, recall of 90%, and F1-score of 58%. So it is possible to observe that the recommender system can perform risk predictions satisfactorily. Therefore, it is promising to assist in decision-making in Scrum-based projects.