|
Bachelorthesis
|
Multi-Objective Optimization for Team Composition in Project Courses
|
|
Advisor
|
Dora Dzvonyar
|
| Author | Teodora Dobos |
| Date | 03/2020 |
| The goal of this thesis was to evaluate multi-objective optimization methods for composing teams in multi-project courses. In order to achieve this goal, we selected three multi-objective heuristic algorithms and applied them to solve the team assignment problem in a concrete past instance of the iPraktikum. We evaluated the applicability, usability and efficiency of these algorithms in a controlled experiment in which we considered performance and assignment quality criteria. For each algorithm, we identified optimal parameter configurations. We compared the solutions generated by the multi-objective methods with the solution computed using the single-objective approach implemented in TEASE. The results show that, among the analyzed multi-objective methods, the MOPSO algorithm calculates the best assignment solution with respect to the metrics that we defined. In addition, compared to a single-objective technique, a multi-objective approach is superior for concrete scenarios of the team assignment problem (e.g. if the student experience level is modeled using fuzzy logic), but it computes assignment solutions that have lower priority objective values. |
