Course Scheduling Using Genetic Algorithms Enhanced by Linear Regression for Data Mining Course Participants

Course scheduling at the beginning of each semester is an absolute must, considering changes in course instructors, changes in the availability of lecture schedules, changes in lecture infrastructure in terms of number, capacity, and time of use, changes in the number of lecture participants, both new and repeat participants. This research aims to design an optimal scheduling system by considering scheduling constraints to avoid conflicts and increase the effectiveness of lecture scheduling management. The linear regression method is used to predict the number of lecture participants using the 2019-2022 academic year data as model data and the 2023 academic year data as testing data to validate the prediction data. Lecture scheduling uses a Genetic Algorithm with a fitness function for the number of cross-class schedules - contracted by repeating students - that conflict with the chromosomes used by courses, classes, lecturers, rooms, schedules, and others. The designed scheduling system has a prediction model with high accuracy and a coefficient of determination (R-Sq.) above 95% and RMSE below 10. This scheduling system is efficient, minimizing scheduling conflicts to 0 percent