Course Timetabling using Genetic Algorithm and Fuzzy Cross-Over

Course timetabling at universities presents a complex problem due to the limited timeframe to create schedules that avoid conflicts between activities. This issue becomes more challenging as the number of activities increases while the available rooms remain constant. Numerous studies have attempted to automate the scheduling process, but their success is often limited to specific cases, meaning their effectiveness may not be applicable in different institutions. One method that has shown potential in solving timetable problems is the genetic algorithm, either as a standalone approach or combined with other techniques. Despite criticisms regarding computational time complexity, genetic algorithms serve as practical global optimization tools, making them suitable for timetabling when computational time constraints are manageable. A hybrid Genetic Algorithm combined with Fuzzy Partitioning is essential for determining the crossover point, one of the key operators in genetic algorithms. In this study, we use a hybrid genetic algorithm with fuzzy crossover to address the course timetabling problem at Pancasila University, focusing on two departments, Informatics and Electro, which share classrooms on the same floor. In this study, we use data from 31 courses; our experiment achieved convergence at generation 78, with a fitness function score of zero, indicating the complete elimination of scheduling conflicts. For further improvement, adjustments could be made to the fitness function to penalize inefficient room usage, reducing the total number of generations to decrease execution time without compromising solution quality, and reducing the mutation rate to enhance solution stability.