A MULTI-OBJECTIVE APPROACH TO THE CLASSROOM ALLOCATION PROBLEM

The present work aims to evaluate algorithms designed to solve the Classroom Allocation Problem (CAP) in university courses. This problem involves assigning class groups with predefined schedules to classrooms distributed across various buildings, considering the university's regulations. Two algorithms were assessed in this study: (i) the Non-Dominated Sorting Genetic Algorithm (NSGA-II) and (ii) the Multi-Objective Late Acceptance (MOLA) algorithm. The solution space exploration in the MOLA algorithm is performed through five neighborhood structures. The algorithms were evaluated using real data based on information from the Institute of Exact and Biological Sciences of Federal University of Ouro Preto. Additionally, both algorithms were compared against an exact solution built through the ϵ-constraint strategy, with an average difference of 45.7% using MOLA and 28.5% using NSGA-II.