Automating Project Team Formation with Heterogeneous Project Preferences and Skill Mix Constraints

It is well established that design team composition is a significant driver of both team member engagement and design project success. Off-the-shelf solutions are available for automating the process of forming design teams with balanced skill mix and diversity profiles. These systems have been successfully applied both at our institution and others to improve student engagement and team performance in design projects with homogenous project opportunities and skill mix requirements (i.e. when forming multiple teams of similar composition to address a common design challenge). However, it has proven challenging to apply available off-the-shelf solutions to form design teams in our Capstone Design program due to heterogeneity in both project opportunities (each project team addresses a unique design challenge) and required skill mixes (which vary by project opportunity). In this paper, we discuss the development and application of a heuristic optimization algorithm that extends the capabilities of existing off-the-shelf project team formation systems, which effectively balance team sizes and diversity profiles within teams, to incorporate students' individual project preferences and project-specific skill mix constraints. Comparing results of applying this algorithm for Capstone Design project team formation with results of applying off-the-shelf solutions for team formation in other design courses with homogenous project opportunities and skill mix requirements, we find that our algorithm satisfies project-specific skill mix constraints and effectively matches students to preferred project opportunities (consistently matching more than 5/6 of students on one of their Top 3 project choices) while offering comparable performance to off-the-shelf project team formation systems in terms of balancing team sizes and diversity profiles within teams. Consistent with findings in prior research, our data indicate that balancing the sizes and diversity of project teams favorably impacts team performance (measured by scores on course assignments) and student engagement (measured by peer evaluations). We are seeking additional insights into these phenomena by incorporating new descriptive variables such as students' project preferences in analyses of Capstone Design project team performance.