Abstract: An efective timetable is crucial or the satisaction o enormous requirement and the ecient utilization o human and space resources, which make it an optimization problem. Traditionally, Traditionally, the problem is solved solved manually by hit and trial method, where a valid solution is not guaranteed. ven ve n i a valid solution is ound it is likely to miss or better solution. These uncertantities have motivated or the scienti!c study o problem, and to develop an automated solution technique or it. The problem is heavily studied or last more than our decades but a general solution technique is yet to be ormulated. Timetabling problem problem can be de!ned to be the problem problem o arranging a number o events into a limited number o timetabling as ollows" #Timetabling is the allocation o sub$ect to constraints that are usually divided into two categories hard and sot. %or achieving the problem statement, in this paper we are going to implement a genetic algorithm. &enetic algorithm is the algorithm which generates a various possible solution and !nally selects the one which the best amongst those solution. Timetabling problems problems are mainly mainly classi!ed as constraint satisaction problems where the main goal is to satisy all problem constraints rather than optimizing a number o ob$ectives. 'cheduling is the arrangement o entities (people, students, lectures, and e)am* into a pattern in space time in such a way that constraints are satis!ed and certain goals are achieved.
Hard Constraints )amples o such constraints are" +o lecturer should have diferent classes at the same time slot. ere cannot be more than - classes or a sub$ect on one day %or each time period there should be sucient resources (e.g. rooms and ecturers* available or all the events that have been scheduled or that time period. •
• •
Soft Constraints 'ot constraints are those that are desirable but not absolutely essential. /n real0world situations it is, o course, usually impossible to satisy all sot constraints. )amples o sot constraints (in both e)am and course timetabling* are" very staf should get at least one !rst hour ecturer having two theory sub$ects has no lab assignments ecturer having one theory may get two lab classes. A particular class may need to be scheduled in a particular time period. ab 1lasses may not be in consecutive hours. ecturers may preer to have all their lectures in a number o days and to have a number o lecture0ree days. • • • •
• •
The proposed system has, 23 no o lecturers, 2'3 no o sub$ects and 213 no o classes per sub$ect per week. ach day has 23 no o hours hour s and we have !ve working days per week. The total no o time slots 4 56.
The problem then becomes assigning 2'613 2'613 number o classes in the 2563 time slots. %or e)ample in the department o 1omputer 'cience and ngineering, 7&/T, there are twelve lecturers and -8 sub$ects with 8 classes per sub$ect per week giving a total number o 9: classes per week. ach day has si) hours and !ve days per week giving ;-< time slots.
Genetic Algorithm: &enetic algorithms (&As* are evolutionary algorithms that use the principle o natural selection to evolve a set o solutions toward an optimum solution. &As are not only very powerul, but are also very easy to use as most o the work can be encapsulated into a single component, requiring users only to de!ne a !tness unction that is used to determine how #good= a particular solution is relative to other solutions.
Steps of Genetic Algorithm: /nitializing" The method o selecting a solution among a set o solution. >reeding" 1ombining properties o two solutions into a single solution. ?utating" 1hanging o sequence solution. 1hoosing and @illing" 'electing a solution and killing i it doesn3t satis!es the !tness unction %or e)ample, hen designing a weekly budget, the amount spent on each item could be stored as a number in a column. This can be thought o as not $ust a list o values but a string o genes. The value in the !rst row might represent the amount o money to spend on rice, and the second row might be the amount o money to spend on caviar and so on. ach o these values might be converted rom base ;< to base - to create a !)ed width binary number. number. ence the problem o minimizing your budget while maintaining your survival is translated into a genetic representation.
A collection o possible budgets could be thus encoded, producing a population o >udget creatures. 7andom populations are almost always e)tremely un!t. /n order to determine which are !tter than others, each creature must be evaluated. /n order to evaluate a creature, some knowledge must be known about the environment in which it survives. ere, environment can be a !tness unction.
