rayjasson98/hybrid-genetic-algorithm-simulated-annealing-for-presentation-scheduling

This repository demonstrates how the Presentation Scheduling problem, which is analogous to the famous University Course Timetabling Problem (UCTP), can be solved using the Hybrid Genetic Algorithm-Simulated Annealing (HGASA) algorithm.

hybrid-genetic-algorithm-simulated-annealing-for-presentation-scheduling's Issues

Slow convergence speed for HGASA algorithm

The improvement of penalty points over iterations is too slow as shown in the sample graph. A better way of changing the scheduling of the presentations in each iteration can be tried. For example, selecting a better neighbourhood structure in the SA phase.

Improvement on meeting SC01 (scheduling of consecutive presentations)

As stated in SC01, the number of consecutive presentations should not exceed and should be as close as possible to a supervisor's preference. Penalty points will be given if the number of consecutive presentations is different from the preferred value. In other words, we are trying to schedule the presentations in a way such that the supervisors can attend most presentations in a row without work overload. As of now, HGASA algorithm can schedule the presentations without exceeding the preferred value. However, it does not work very well to arrange other presentations (those supervised by the same supervisor) consecutively. A better penalty scheme or change in the scheduling method is expected.