Currently, the transversal competency of collaborative teamwork has gained significant importance. Many teaching methodologies rely on distributing the students into different groups, including cooperative learning, project-based learning, problem-based learning, discussion groups, or activity stations.

For these methodologies to be effective, the group must function properly. Therefore, there is a need to examine how students are distributed into groups.

There are primarily three approaches:
1. The teacher manually forms groups based on specific criteria.
2. Students self-select their groups.
3. Groups are formed randomly.

Studies suggest that the first approach is preferable since groups are more task oriented than self-selected groups [1]. Also, self-selected groups often result in stronger students clustering together dividing tasks among themselves, while weaker groups fail to reinforce or may reinforce misconceptions [2]. Random grouping, on the other hand, risks creating imbalanced groups. The first approach, however, imposes a significant burden on teachers, especially in large classes or when multiple criteria must be considered, making the task nearly unmanageable without computational algorithms.

This paper presents an updated version of a previously published algorithm [3] designed to automatically group students using complex network science. In this algorithm, students are represented as nodes within a network, connected based on their similarity across various criteria. The algorithm is flexible enough to allow the user to specify the size of the groups and determine the number of criteria to be considered.

The updated version incorporates the option to include student preferences, that has some advantages as recent studies such as [4] highlights: students tend to prefer grouping based on their social networks, which leads to increased engagement in group tasks.

The algorithm was applied to a cohort of aviation university students when working in a group project in teams of three for the Principles of Flight I course. Students were grouped based on their preferences and their grades in the Basic Principles of Physics course.

The results demonstrate that most groups were academically heterogeneous, and most preferences were respected.

References:
[1] Hassaskhah, J., & Mozaffari, H. (2015). The impact of group formation method (student-selected vs. teacher-assigned) on group dynamics and group outcome in EFL creative writing. Journal of Language Teaching and Research, 6(1), 147.
[2] Oakley, B., Felder, R. M., Brent, R., & Elhajj, I. (2004). Turning student groups into effective teams. Journal of student centered learning, 2(1), 9-34.
[3] Vallès-Català, T., & Palau, R. (2023). Minimum Entropy Collaborative Groupings: A tool for an automatic heterogeneous learning group formation. Plos one, 18(3), e0280604.
[4] Chen, R., & Gong, J. (2018). Can self selection create high-performing teams?. Journal of Economic Behavior & Organization, 148, 20-33.

Keywords: Collaborative groups, group formation, automatic group formation, self-selection.