Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Applied Mathematical Sciences, Vol. 9, 2015, no. 130, 6449 - 6460
HIKARI Ltd, www.m-hikari.com
http://dx.doi.org/10.12988/ams.2015.53292
Quadratic Assignment Approach for
Optimization of Examination Scheduling
Fadhilah Ahmad*, Zarina Mohammad, Hasni Hassan,
Ahmad Nazari Mohd Rose and Danlami Muktar
Faculty of Informatics and Computing
Sultan Zainal Abidin University (UniSZA)
Tembila Campus, 22200 Besut, Terengganu, Malaysia *Corresponding author
Copyright © 2015 Fadhilah Ahmad et al. This article is distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.
Abstract
The increase in student enrolment into institution of higher learning, and the great
increase in flexibility in choice of courses has made examination scheduling very
difficult to handle in the traditional scheduling system. This paper investigates a
quadratic assignment approach in examination scheduling by presenting a
quadratic assignment approach to eliminate or reduce the level of conflicts in the
examination timetabling. The examination with the highest number of students
will be prioritized across all faculty courses, followed by other criteria such as
courses with the highest number of students from different faculties. A framework
for optimizing and handling the conflicts at all constraint levels is proposed to
solve the examination scheduling problem.
Keywords: Optimization, Quadratic Assignment Problem, Examination scheduling
1. Introduction
Examination scheduling is the allocation of examinations subject while
considering the constraints of such as examination period, time slots, and rooms
sizes to entities being placed in space time when fulfilling a set of desirable
purposes or objectives. Examination scheduling problems involves time-related
scheduling and combinatorial optimization that tends to be unraveled with
combination of search and heuristics, which usually lead to satisfactory but
solutions that are not fully optimal.
6450 Fadhilah Ahmad et al.
In timetable scheduling problems, events must be slotted to certain times which
satisfy several constraints. It is difficult to design knowledge-based and operation
research based algorithms to solve such problems due to the complicity of the
system. On the other hand, constraint satisfaction techniques have been used to
solve hard constraints; however, it is more difficult to handle soft constraints such
as that of the examination scheduling problem.
These sets of objective demand an automated system built around mathematical
models or algorithm solutions for constraints satisfaction. Such constraints may
involve students having more than one examination at the same time or students
taking more than one course within a day, under-utilization of room facilities and
ensure capacity of individual rooms is not exceeded at any time throughout the
examination session, exceeding duration of examination periods and inadequate
availability of lecturers.
The Quadratic Assignment Problem (QAP) is a well-known combinatorial
optimization problem with a wide variety of practical applications. Many
heuristics solution has been proposed for QAP. This paper presents an
improvement of the quadratic assignment problem initiated in [1].
2. Problem definition and related works
An improper examination timetable may hinder the students for better preparation
for the examination. In addition, there are lots of subjects involved in the
examinations to be scheduled, a large number of students who are registering
different subjects, the limited number of rooms or exam halls that are available,
and some other constraints related to individual student’s examination timetable.
That is, a student is not scheduled to sit for more than one subjects at the same
time. The scheduling problem created by these set of circumstances clearly poses
an interesting intellectual challenge. Therefore, this study is to explore this
challenge and hence, will propose a solution to this problem.
3. Optimization approaches for examination scheduling
Examination timetabling is a well-researched and important area [3]. There is
some novel research that focused on the modeling and formulation of
Examination Timetabling Problem (ETP), and a number of research in the
applications of heuristics to solve the problem as shown in Table 1.
A new binary model for examination time tabling problem has been investigated
to deal with same causes that are scheduled in different time slots. It is also
possible to schedule two or more courses in a room simultaneously. The model
also considers some hard and soft constraints, where the objective function is set
in such a way that soft constraints are satisfied as much as possible [2].
Quadratic assignment approach for optimization 6451
Likewise in a research by [10], an integer programming (IP) model was adopted
to address certain constraints of examination scheduling problem such as
restricted availability of classrooms and the increased flexibility of the student’s
choices of courses. It assigns courses to the available time slots and rooms. The
quality of the schedule produced depends on the relative position of the courses
assigned to the available time periods, a condition that the IP model attempts to
satisfy by constructing groups of courses that are assigned to groups of time
periods. The system was adopted by the Athens University of Economics and
Business for use and with success. The approach is said to be theoretical
development and has not analyze an actual computer-based implementation
The issue of assigning invigilators to examination which are faculty lecturers or
graduate students in educational institutions using a computer based system has
been addressed by [11]. They have adopted Analytic Network Process model for
estimating exam objective weights. The exam weights used arbitrary assignment
ratio scale numbers to assign invigilators for certain examination slot.
An Ant colony optimization algorithms was adopted by [18] in solving
examination scheduling problem which was described as combinatorial
optimization problems derived from the field of swarm intelligence. The research
described how real ant colonies find shortest paths between food sources and their
nest. Ant Systems (AS) was implemented in the algorithms for solving the
traveling salesperson problem. The solution approach consists of n cycles of
examination construction process. Each cycle consists of m ants where each m ant
constructs a viable solution. Each ant builds a complete tour that visits all nodes.
Honey-bee mating optimization algorithm approach was pioneered by [19] to
address examination timetabling. This process was described as a swarm-based
approach to optimization inspired by real honey-bee mating. The algorithm
mimics the mating flight of a queen bee. The mating process (i.e. generating a
new solution) of the queen (current best solution) begins when the queen leaves
the nest to perform a mating flight during which time the drones (trial solutions)
follow the queen and mate with her. The algorithms start by initializing the
population and the best solution is selected as the queen.
A quadratic assignment was adopted in 2014 by [1] to address examination
timetabling problem by developing a mathematical model. The model will
consider E exams needed to be scheduled in a limited number of T, time slots and
limited number of R, room and capacity, taking into account all examination must
be scheduled within the time frame. The room capacities and the restrictions that
students cannot take more than one examinations at the same time are among
other constraints. The heuristic solution is formed to identify possible constraints
6452 Fadhilah Ahmad et al.
and groups them as hard and soft constraints. The heuristic algorithm ensure the
solution is improved if needed, until an optimal solution is achieved and all
constraints are satisfied.
[21] Has adopted a transformation process to solve examination time tabling as
quadratic assignment problem. The objective was to separate lists of high
interactions as far apart as possible in other to obtain a well-spaced schedule with
the use of activity and traffic cost transformation. A super imposition was
considered to spread courses from the same faculty by interweaving with courses
from different faculties.
Table 1: Optimization algorithms for examination scheduling problems
Bil. Authors date Model Strength Weaknesses
1 Thin-Yin
Leong, Wee-
Yong Yeong
1990[16]
Hierarchical
approach based
on Quadratic
Assignment
problem
Feasible solution, less
computational effort, easy to
understand, implement, and
interact.
Heuristic technique is used
in each stage causing
problems when the size
constraints increase.
2 Thin-Yin
Leong, Wee-
Yong Yeong
2012[21]
Quadratic
Assignment
Problem
Reduce the assignment cost
by
Adoption of monotonic
examination assignment.
Results were not
spectacular, due to the
complicity of the grouping
process.
3 Nasser R.
Sabar, Masri
Ayob Rong
Qu, Graham
Kendall
2011[19]
Honey-Bee
Mating
Optimization
Produce good quality
solutions for benchmarks
ETP.
This solution is yet to
receive wide acceptance.
4 Micheal Eley,
2006 [18]
Ant Systems
Optimization
The solution is capable of
solving large real world
exam timetabling problems.
This approach requires
proper adjusting parameters
to achieve an optimal
solution
5 Dimopoulou
and Miliotis
2001[10]
Integer
Programming
Contributed numerous
improvements to the
decision-making process
Schedule outcome is
parameters dependent.
Quadratic assignment approach for optimization 6453
Table 1: (Continued): Optimization algorithms for examination scheduling
problems
4. Proposed framework
This paper proposes a framework for examination scheduling based on QAP. The
framework consists of five phases (refer to Fig. 1) in which three phases applied
QAP approach. In Phase 1, all parameters for examination scheduling are
determined. Constraint tables are generated in Phase 2. Assignment of parameters
is done in Phase 3. The other two phases are for solving the constraints and
applying the solution. The whole phases are run consecutively, one after the other
from Phase 1 until 5 to obtain an optimized solution for examination scheduling.
The solution can be used to generate examination time table at the end of each
academic semester or whenever it is required.
4.1 Quadratic Assignment Problem
The Quadratic Assignment Problem (QAP) was introduced by Koopmans and
Beckman in 1957 in the context of locating "indivisible economic activities". The
objective of the problem is to assign a set of facilities to a set of locations that will
minimize the total assignment costs. The assignment cost for a pair of facilities is
a function of the flow between the facilities and the distance between the locations
of the facilities [13]. Due to wide assortment of applications, QAP has been
extensively used to formulate other practical problems such as economic problems
[22], Scheduling problems [24], Timetabling problems [21], and Facility layout
problems [23].
6 Alireza
Rashidi
Komijan and
Mehrdad
Nouri Koupaei
2012[2]
Binary Model Eliminates all conflicts in
individual student timetable.
The model dose not satisfy
all soft constraints such as
consecutiveness, and
mixed duration.
7 Sagir and
Ozturk,
2010[11]
Analytic
Network Process
Model
Considers system dynamics;
people's expectations, and
system parameters.
Focus more on the
Prioritization of invigilator
assignment.
6454 Fadhilah Ahmad et al.
This work enhances the QAP to emulate the scheduling of examination time table
such that the objective function will eliminate or reduce the rate of conflict at each
assignment. The enhanced QAP considers the complicity of constraints; conflict,
capacity of individual rooms, capacity of the whole rooms, and consecutiveness of
scheduled exams. The original QAP was implemented by considering only a
limited number of constraints. In order for the QAP to be applied for examination
scheduling, its formula needs to be adjusted so that all considered constraints can
be taken into account. The mathematical formula for this purpose is describe in
the following sections.
4.2 Mathematical Formulation
An E number of exams need to be scheduled in a limited number of T time slots
and limited number of R room and capacity, taking into account all examination
must be scheduled within the time frame and the rooms capacities and the
restrictions that students cannot take more than one examination at the same time
as other constraints.
N= {1, 2, 3 … n}
:n T E R , the set of all permutations.
ijt T , is the time slot for an exam to be assigned to a room
1 1
minn n
ij ij
i j
t
i = 1, 2, 3 … n
j = 1, 2, 3 ... (n-1)
Quadratic assignment approach for optimization 6455
Fig. 1: Proposed framework for Examination scheduling based on Enhanced QAP
Represents the assignment of exams to specific time slot and different rooms,
to prevent conflicts. The conflicts degree of each examination is minimized when
exams i are assigned to room j at a specific time and is denoted by ij ijt where
1 if exam i is assigned to room r
0 otherwise ij
6456 Fadhilah Ahmad et al.
Room Assignment
An examination can be assign to one or more room depending on the number of
the student to take the examination or otherwise, ie two or more examination in a
room. But for all examination assigned to the room/s the number of students to
seat for the examination most be less or equal the seat capacity/ies of the room/s.
this is mathematically express in the figure below.
1
n
i j i j
j
e r se sr
ie E ∀ i = 1, 2 … n
jr R ∀ j = 1, 2 … (n-1)
Sei= Size of student in the exam e
i
Srj = Size of room
jr
Course Assignment to time slot
For each course assignment to time slot it ensure that no one student have more
than one examination for a particular time slot ie for all examination assigned to a
particular time slot, it ensure that all the examination have no common student.
This can be express mathematically in the figure below.
1 2 ...i i ij ne E e t e e e
1
n
i i
i
e e
i ije t
i = 1,2,…n
lei = student in examination i
In examination scheduling, Quadratic assignment reduces the rate of conflicts
during the scheduling of the examination timetable. The examination for a course
taken by students from different faculties is being prioritized. Next, the same
courses with students from the same faculty, but different program is considered
and finally the highest number of repeating student from the previous semesters.
At each stage, course with the highest number of registered students is given
prioritization.
Quadratic assignment approach for optimization 6457
The following algorithm and flowchart describe the scheduling process:
Initialize the parameters
Get the total number of available seats, tot_seats and total number of students per time
slot, tot_students
If tot_seats ≥ tot_students
Begin assignment
{Set the current solution, current_sol = initial_sol
Apply a sequential search to list of examination E
Select an examination E in ascending order of conflicts where E1 > E2 >........En
}
If current solution is best solution, current_sol = best_sol
Apply current_sol to E1
If the result is feasible
Apply current_sol to E1………En
Else {
Improve current_sol
Update the best solution, best _sol = current_sol
}
End
The framework defines all possible conflicts and groups them as hard and soft
constraints. A number of algorithms for solving the conflicts have been proposed
to eliminate the possible constraints in the system.
5. Conclusion and recommendation
In this paper, we highlighted some weaknesses and strength of the related studies
on examination scheduling. The proposed examination timetabling problem was
formulated based a Quadratic Assignment Problem (QAP). Since the number of
examinations can be greater than the number of time periods and in a quest to
minimize all possible conflicts, the objective function is set in such a way that,
hard constraint satisfaction is maximized. The future work of this research is to
implement an examination scheduling system which will allow simulation and
evaluation of various examination data including the benchmark data. The result
of the implementation will provide the planning of examination scheduling
becomes easier and fulfil the user expectation.
Acknowledgements. The author would like to extend appreciation of the support
given by Universiti Sultan Zainal Abidin (UniSZA) and the sponsorship by Kano
State Government.
6458 Fadhilah Ahmad et al.
References
[1] M. Danlami, A. Fadhilah, M. Zarina, H. Hasni, H. Z. Aznida, Examination
Scheduling System based on Quadratic Assignment, Proceeding of the Third
International Conference of Information and Applications, 8-10 October, 2014,
Kuala Terrenganu, Malaysia.
[2] Alireza R. Komijan and Mehrdad N. Koupaei, A new binary model for
university examination timetabling: a case study, Journal of Industrial
Engineering International, 8 (2012), no. 28.
http://dx.doi.org/10.1186/2251-712x-8-28
[3] R. Qu, E. K. Burke, B. McCollum, L. T. G. Merlot, and S. Y. Lee, A survey of
search methodologies and automated system development for examination
timetabling, Journal of Scheduling, 12 (2009), no. 1, 55–89.
http://dx.doi.org/10.1007/s10951-008-0077-5
[4] S. Desroches, G. Laporte and J-M. Rousseau Horex, A Computer Program for
the Construction of Examination Schedules, INFOR, 16 (1978), 294-298.
[5] Aldy Gunawan, Modeling and Heuristic Solutions of University Timetabling
Problems, PhD Thesis, University of Singapore, 2008.
[6] Zaki Salikon, Examination Timetabling Using Genetic Algorithm Case Study:
KUiTTHO, Masters thesis, Universiti Utara Malaysia, 2005.
[7] Nelishia Pillay, An Empirical Study into the Structure of Heuristic
Combinations in an Evolutionary Algorithm Hyper-Heuristic for the Examination
Timetabling Problem, SAICSIT ’10, 2010.
http://dx.doi.org/10.1145/1899503.1899531
[8] Nasser R. Sabar, Masri Ayob, R. Qu and G. Kendall, A graph coloring
constructive hyper-heuristic for examination timetabling problems, Springer
Science + Business Media, LLC, (2011).
[9] G. M. White and P-W. Chan, Towards the Construction of Optimal
Examination Schedules, INFOR, 17 (1979), 219-229.
[10] M. Dimopoulou, & P. Miliotis, Implementation of a university course and
examination timetabling system, European Journal of Operational Research, 130
(2001), 202–213. http://dx.doi.org/10.1016/s0377-2217(00)00052-7
[11] Mujgan Sagir, Zehra Kamisli Ozturk, Exam scheduling: Mathematical
modeling and parameter estimation with the Analytic Network Process approach,
Mathematical and Computer Modelling, 52 (2010), no. 5–6, 930–941.
Quadratic assignment approach for optimization 6459
http://dx.doi.org/10.1016/j.mcm.2010.05.029
[12] Salem M. Al-Yakoob, Hanif D. Sherali, Mona Al-Jazzaf, A mixed-integer
mathematical modeling approach to exam timetabling, Comput. Manag. Sci., 7
(2010), 19–46. http://dx.doi.org/10.1007/s10287-007-0066-8
[13] T. C. Koopmans, M. J. Beckmann, Assigment problems and the location of
economic activities, Econometrica, 25 (1957), 53-76.
http://dx.doi.org/10.2307/1907742
[14] Andrew L, Ang J. Chin, Ho W. Kit, and Oon W. Chong, A Campus-Wide
University Examination Timetabling Application, American Association for
Artificial Intelligence, 2-3, 2000.
[15] M. Tomas, Real-life Examination Timetabling, Space Management and
Academic Scheduling, Purdue University, 1-2, 2013.
[16] Thin-Yin Leong, Wee-Yong Yeong, A Hierarchical Decision Support
System for University Examination Scheduling, National University of Singapore,
4-10, 1990.
[17] Barry McCollum, Paul McMullum, Andrew J. Parkes, Edmund K. Burke,
and Rong Qu, New Model for Automated Examination Timetabling, Annals of
Operation Research, 194 (2012), no. 1. 291-315.
http://dx.doi.org/10.1007/s10479-011-0997-x
[18] Micheal Eley., Ant Algorithms for the exam timetabling problem, E. K.
Burke, H. Rudová (Eds.): PATAT, (2006), 167–180. ISBN 80-210-3726-1.
[19] Nasser R. Sabar, Masri Ayob, Graham Kendall., Solving Examination
Timetabling Problems Using Honey-bee Mating Optimization [ETP-HBMO],
Multidisplinary conference on the scheduling: theory and application (MISTA),
(2009), Dublin Ireland.
[20] Miranda J., Pablo A. R, Robles J. M., A Web architecture based decision
support system for course and classroom scheduling, Decision Support Systems,
52 (2012), 505–513.
[21] Thin-Yin L., and Wee-Yong Y., Examination Scheduling- A Quadratic
Assignment Perspective, International Conference on Optimisation: Techniques
and Applications, Singapore, 1987.
[22] D. R. Heffley Decomposition of the Koopmans-Beckman Problem,
Regional Science and Urban Economics, 10 (1980), no. 4, 571-580.
6460 Fadhilah Ahmad et al.
http://dx.doi.org/10.1016/0166-0462(80)90018-6
[24] Mostafa Kazemi, Saeed Poormoaied, and Ghasem, Eslami, Optimizing
combination of job shop scheduling and quadratic assignment problem through
multi-objective decision making approach, Management Science Letters, 2
(2012), no. 6, 2011-2018. http://dx.doi.org/10.5267/j.msl.2012.06.020
Received: April 15, 2015; Published: November 2, 2015