A New Multiple Objective Cuckoo Search for University Course Timetabling Problem

Abstract

University course timetabling problem (UCTP) is classified into combinatorial optimisation problems involving many criteria to be considered. Due to many conflict objectives or difference objective units, combining conflicting criteria into a single objective (weight sum approach) may not be the best way of optimisation. The UCTP is well known to be Non-deterministic Polynomial (NP)-hard problem, in which the amount of computational time required to find the solution increases exponentially with problem size. Solving the UCTP manually with/without course timetabling tool is extremely difficult and time consuming. A new multiple objective cuckoo search based timetabling (MOCST) tool has been developed in order to solve the multiple objective UCTP. The cuckoo search via Levy flight (CSLF) and cuckoo search via Gaussian random walk (CSGRW) using the Pareto dominance approach were embedded in the MOCST program for determining the set of non-dominated solutions. Eleven datasets obtained from Naresuan University in Thailand were conducted in computational experiment. It was found that the CSLF outperformed the CSGRW for almost all datasets whilst the computational times required by the proposed methods were slightly difference.