SOLVING UNIVERSITY COURSE TIMETABLING PROBLEM USING GENETIC ALGORITHM Master Thesis

This is to certify that we have read this thesis and that we find it fully adequate in scope, quality and content, as a thesis for the degree of Master of Science. continously asking me " how is the thesis going? " , because she encouraged me to persevere without knowing it. Also I would like to thank to my best friend, my colleague, my dear husband Tarkan Aydın for his understanding, patience, encouragement, love and support. Also I thank him for his helpful discussions. The university course timetabling problem consists of allocating a number of courses to a limited set of resources such as rooms, timeslots, set of lecturers and group of students in such a way to satisfy predefined constraints. The constraints can be divided into two groups: hard constraints and soft constraints. A timetable has to satisfy all hard constraints in order to be feasible and it should satisfy as much as possible soft constraints in order to be good quality. The university timetabling problem is in class of NP-hard problems. This means that the amount of time and work required solving this type of problems increases exponentially with the problem size. This makes these problems more difficult and time consuming. Therefore optimization techniques are used to solve them and produce optimal or near optimal feasible solutions instead of exact solutions. Genetic algorithms are considered as an efficient approach for solving this type of problems. Genetic algorithm is based on the principle of evolution first described by Charles Darwin and it tries to mimic some features of nature such as selection, crossover, mutation and replacement. Through these probabilistic operators, genetic algorithms choose the optimal solution among a set of alternate solutions which compete with each other. In this thesis, first a general description of genetic algorithms and theoretical background, Schema Theorem, which describes the reasons of efficiency of genetic algorithms are given. Then a real life university course timetabling problem is examined with data coming from Bahcesehir University, Faculty of Arts and iv Sciences. A solution based on genetic algorithm is proposed for the problem. The genetic algorithm implementations are described and applied to the sample problem. Moreover, determination of the best parameter values, such as the population size, mutation and crossover rate, etc. is tried. We work on this problem by introducing modified genetic operators which do not allow violations of any hard constraints, as a …