Team formation in engineering classrooms using multi-criteria optimization with genetic algorithms

• Main Authors: Agarwal, Jutshi, Piatt, Emily, Imbrie, P. K.
• Format: Conference Proceeding
• Language: English
• Subjects: Behavioral sciences; genetic algorithm; Genetics; Industries; Leadership; multi-criteria optimization; Question answering (information retrieval); Task analysis; team formation; Teamwork
• ISSN: 2377-634X
• DOI: 10.1109/FIE56618.2022.9962741

The research-to-practice paper presents applications of genetic algorithms using multi-criterion optimization to designate student-teams in an academic setting. Teamwork skills are becoming a more desirable trait in industry today and hence more instructors are using teamwork in their engineering courses. ABET also requires engineering degree programs to have student outcomes that provide them with "an ability to function effectively on a team whose members together provide leadership, create a collaborative and inclusive environment, establish goals, plan tasks, and meet objectives." This, however, comes with its challenges because the literature suggests that random or student-selected teams tend to lead to dysfunctional behaviors. Instructor-designated academic teams based on skills, learning personalities, and demographics are known to be more effective in promoting learning and positive interactions. Creating optimal homogeneous or heterogeneous teams based on multiple criteria (e.g., prior knowledge, skills, abilities, psychosocial, demographics) can be a complex and time-consuming task when done manually, especially when large numbers of students are involved.This study presents an expansion of previous work that used single-criterion genetic algorithm optimization of teams in a first-year classroom. The research question answered in this study is, how do teams formed using an algorithm that optimizes multiple criteria with genetic algorithms represent heterogeneity when compared to teams formed manually? Using a Vector Evaluated Genetic Algorithm (VEGA), optimization of multiple skills is conducted to attain uniform heterogeneity in the classroom. The algorithm uses discrete integer-type representations as a basic unit of team configuration. Self-reported student competency data on three different computational skills were used. Teams were formed for approximately 1300 students enrolled in a first-year engineering design thinking course at a large Midwestern University. Four-member teams were maximized with a minimum number of teams with 3 students in each section. Average skills of the teams are calculated and the standard deviation in class is minimized for each of three skills parallelly. The algorithm also aims to maintain diversity of teams based on gender and ethnicity.Results presented include visualization of team-configurations and comparison with teams formed manually using the same criteria. The aim of the algorithm is to optimize studen