AI-driven decision support for SCRUM team selection in smart city software development projects

Abstract

Objectives: This study proposes a machine learning-enabled application designed to support project management in selecting optimal SCRUM development teams for smart city software projects. These projects exhibit significant variability in complexity, ranging from localized initiatives to large-scale government-level applications. The solution aims to enhance decision-making by aligning team configurations with project goals, complexity, and resource constraints. Prior work: Building on established SCRUM and SAFe frameworks, the research leverages prior studies analyzing role distribution within development teams across four scenarios. These scenarios range from minimal setups with a single Business Analyst (BA) to complex configurations involving a Product Owner (PO) and a Product Manager (PM). The work addresses the challenges of adapting team roles to diverse project demands, particularly in the context of smart cities. Approach: The research employs a case-study methodology, combining quantitative data from more than 50 professionals across more than 25 companies and qualitative interviews with three experts. A machine learning model incorporates these empirical insights to recommend team structures tailored to project-specific characteristics and historical success metrics. Results: The application accurately identifies the most suitable SCRUM team configurations for diverse project criteria. Validation results indicate improved outcomes, such as reduced development time, enhanced team productivity, and better alignment with stakeholder expectations. Implications: This research provides academics and practitioners with tools to systematically optimize SCRUM team structures. It addresses the unique complexities of smart city projects, offering scalable solutions for various levels of project intricacy. Value: By integrating AI-driven insights with practical SCRUM principles, this study delivers a novel approach to managing smart city software projects. It bridges the gap between theoretical frameworks and the dynamic demands of real-world applications, ensuring scalability and efficiency.