Tabu Search with Explicit Adaptive Guiding Heuristic for the Examination Timetabling Problem

Abstract

Tabu Search algorithm (TS) is one of the leading methods for solving the Examination Timetabling Problem (ETP). The challenge with TS which uses only one solution in searching is that it can easily miss some areas of the search space and chose the best solution based on the current search step and position without taking the future steps into consideration. To address these problems, an explicit and adaptive guiding heuristic was proposed for the standard TS algorithm with the goal of enhancing its performance. The standard TS and the TS with the explicit adaptive guiding mechanism, the TS_G algorithm were implement using the Java programing language (version 8.92) with the MySQL database and experimented with data from Bells University of Technology, Ota. Performance of the two algorithms were evaluated using first Order Conflict Counts (OCC), Second OCC and third OCC for students and invigilators respectively, in addition to memory consumption. The TS and TS_G yielded an average first OCC 0.0 and 0.0 students and 0.0 and 0.0 for invigilators. Similarly, TS and TS_G yielded average second OCC of 18.8 and 10.5 for students and, 0.0 and 0.0 for invigilators respectively. Furthermore, TS and TS_G yielded average third OCC of 33375.0 and 9922.1 for students and 0.0 and 0.0 for invigilators, respectively. The TS_G algorithm outperformed the standard TS algorithm with lower conflicts violations. Both algorithms however recorded similar memory consumption.