Meta-heuristic approaches to tackle Skill Based Group allocation of Students in Project Based Learning Courses

In the arena of software engineering, Project Based Learning (PBL) is one of the fundamental components of practical based assessment. PBL involves team formation where necessary skills are needed to execute the project. Traditionally, the teams were randomly allocated based on individual preferences. To cab on this issue, preference based model needs few refinements such as skills needs to be identified by the facilitator while the students provide the necessary skill data. This way, students get assigned based on their skill rather than just random allocation. In a worst case scenario for random allocation, a team can end up with a very strong team having high skills or vice versa where a team has all of its members with limited skill or few skills are missing. The group created by skill preference would allow each group to more or less have the same strength and nearly all skills would be present in a group. In this paper, a method is extended from its original to cater for other state-of-the-art optimization techniques rather than just genetic algorithm to find a method that can suit small or large dataset. The objective function takes into account the differences between the total skill set of each group with the average total skill set needed for each group and the missing skill penalty of each group is added. Missing skill penalty is incurred due to not satisfying all the constraints such as non-presence of all the skills in a group. The skill rating allows better selection of members in a software engineering course. The results discussed in this paper are from 5 courses of one university.