Optimization of Study Plan for Irregular Students: Genetic Algorithm and Mixed Integer Programming

Academic irregularity in engineering courses presents a complex challenge of resource allocation and time planning. This work proposes and compares two methods of curriculum optimization from the perspective of the individual student: one based on Evolutionary Computation (Genetic Algorithm, GA) and an exact model of Mixed Integer Linear Programming (MIP). The curriculum structure is modeled as an Acyclic Directed Graph (AGD), from which a topological criticality metric used by both methods to prioritize strategic disciplines is derived. Constraints of dynamic prerequisites, schedule availability, and semester workload are incorporated consistently across the two models, allowing for direct comparison under the same semester decision cadence. Experiments with real data from the Systems Engineering course at UFMG, in a protocol of 30 independent executions of the AG, show that the exact model solves each semester the proven optimality in less than a second, while the AG delivers solutions of equivalent quality with stable execution time and no dependence on external solver.