Utilization of Timetable Management System To A Medium Scaled University
Utilization of Timetable Management System To A Medium Scaled University
Utilization of Timetable Management System To A Medium Scaled University
net/publication/320708066
CITATIONS READS
2 12,064
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Saminda Premaratne on 30 October 2017.
ABSTRACT : University timetable construction is hardworking and complicated task when there are large
number of course arrays and limited resources. As a result, universities and some institutes tend to solve this
issue manually even; the results may not always fully optimal. In this paper, we discuss about a framework of
utilizing timetable management system to a medium scale university for resource optimization. Our endeavor
through the overall research was to develop an automated timetable management system to the faculty of IT at
university of Moratuwa to overcome the mentioned scheduling issues. We conducted a preliminary study and
hypothesized it can be achieved by using Genetic Algorithm. In the solution, each individual called chromosome
and it was evaluated using a fitness function in the implementation process. Five great Chromosomes with
higher fitness value considered as optimal solution or timetable schedules. The timetable administrator can
further refine the most suitable timetable. Tools such as PHP, Yii with MVC architecture and MYSQL were used
in this system. Finally, this system was tested and evaluated in the university background and we suggest this
framework is more desirable for medium scale universities.
KEYWORDS -Fitness, Genetic Algorithm, medium scale, optimal, Timetable Management System, utilize
www.ijmret.org Page 8
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
www.ijmret.org Page 9
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
www.ijmret.org Page 10
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
www.ijmret.org Page 11
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
Lecturer Overlap – One lecturer can’t be in two 3.3.3 Database Design of Timetable
rooms at the same time Management System
Student Overlap - One student can’t be in two The Timetable Management System
rooms at the same time Database abbreviated as ttms. It stores data of
students, lecturers, users, degree programs, subjects,
3.3 Design of the TMSFIT timetables and some more. Student data, resources
We will be discussing here, what the data, lecturer data, batch data, subject data and
TMSFIT does and what are the relationships among timetable data which can retrieve from the database.
each module or level. Admin has the authority of modifying and deleting
The system design was categorized as First level, data. Student, lecturer and course details were taken
Second level and Third level as shown in Top level from the faculty of Information Technology at
design diagram in Fig. 3as below. University of Moratuwa.
Timetable 3rd
Screen View Timetabl Level
e
Download
Print
Web Site
Timeta
Fig.3. TopbleLevel Design Diagram
View
www.ijmret.org Page 12
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
analysis was conducted at each different user That CURD generator was important to generate
category getting help of admin of the current controller and views.
timetable management system.
After the system study, the Software Requirement 3.4.3 Logic Implementation
Specification (SRS) for the proposed system was Implementation of the logic of the timetable
prepared. (Algorithm), constraints or rules, verification,
timetable generation, view, delete and edit
3.4.1 Interface Implementation operations were in this second level module.Yii used
There were fifty interfaces included in this AlgoritmController class and Algorithm class for the
TMSFIT system. Those classes are residing in model algorithm development. Hard constraints which
and controller of the MVC (Model, View and can’t be violated were reside in the calFiness() and
Controller) architecture.Some of the classeswere always use to evaluate the fitness value of the
associated with particular interfaces as follows. timetable schedule.
LoginForm, SiteController and User classes
used for login interface. 3.4.3.1 Genetic Algorithm Implementation
DashboardController class used for dashboard In the initialization process of the GA,
interface as Fig. 5. Chromosome or a class schedule must be defined
UserController class used all the interfaces with first. Eg public $_chromosomes. Then, initial
CRUD operators. population was created and it is usually randomly
StudentController, Student and generated 100 chromosomes as the gene or pool.In
Studentenrolsubject classes used for Create the evaluation process of the GA used to, find better
Student, Update Student, Delete Student, individuals in each generation using fitness function
Manage Student, View Student Timetable as the main goal. The fitness value was calculated by
calFitness()how well it fits with our desired
3.4.2 Database Implementation requirements. The main operations of GA such as
MYSQL was used as the back end of this selection, crossover and mutation were evaluated
system. Because it includes number of engines and against by fitness function.
delivers SQL commands to operate database.As the Chromosome as Fig.5.or Schedule Evaluation done
first step of developing this system, a proper with calFitness(). If no room overlapping then
database was constructed in phpMyAdmin. In this increase the score by 1
system, it was called ttms.InnoDB engine use for If it has enough room space then increase the score
foreign keys, support transactions and row level by 1
locking.According to the ER diagram in design, with If required resources are there, then increase score
main entities such as student, subject, timeslot, by 1
lecturer, degree and department main tables were Check lecturer overlapping then increase score by 1
generated. Moreover, batch, degree, department, Check student overlapping then increase score by 1
employee, employeesubject, preferredtime, resource, Finally score of criteria should be $ci += 5;
student, studentenrolsubject, subject, timeslot, FunctioninitObject($numberOfCrossoverPoints,$mutatio
timetabletimeslot and user tables also nSize,$crossoverProbability,$mutationProbability,$fitnes
s,$subjectClass)
implemented.Primary keys were assigned properly
{
to avoid duplicate fields.In addition, user also can // reserve space for time-space
back up database, import database and export slots in chromosomes code
database any time. $this->_slots = new
SplFixedArray(Schedule::DAYS_NUM *
Then,Process of Mapping Database tables
Schedule::DAY_HOURS * count(Resource::model()-
with Model class were done. In that case,we logged >findAll()));
on toYii code generator and used the model $this->_criteria = newSplFixedArray( 5 *
generator, which generates a model class for the count($subjectClass));
$this->_mutationSize = $mutationSize;
specified database table Eg batch. Then, it generated
$this->_numberOfCrossoverPoints =
all the appropriate user interfaces mapping with $numberOfCrossoverPoints;
tables of the database. Later, using CURD generator, $this->_crossoverProbability = $crossoverProbability;
we could generate a view script file which displays a $this->_mutationProbability = $mutationProbability;
Fig.5.Code
$this->_fitnesssegment
= $fitness;of Chromosome
form to collect input for the specified model class.
$this->_subjectClass = $subjectClass;
$labs = Resource::model()->findAll("type = 'lab'");
www.ijmret.org Page 13
$lHalls = Resource::model()->findAll("type != 'lab'");
$this->_allClassRoom = array_merge($lHalls,$labs);
$this->_noOfClassRoom = count($this-
>_allClassRoom);
$this->_noOfLabs = count($labs);
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
3.4.3.2 Selection, Crossover and Mutation Therefore, admin has to manually changethe
operations schedule to make it more accurate.Fig.6. will show
Selection chooses superior individuals in the automate TMSFIT with constraints to be
every generation. This discards the bad designs and changed.
keeps only the best individuals in the population.
Use 100 Chromosomes as this pool (gene). Then, As discussed, timetabling system
calculate fitness value for chromosome by using administrator has all the responsibilities of the
calFitness() and that function uses the hard Timetable Management System including manual
constraints and increase the score one by one 0 to 5. changes to the generated timetable. Further, he or
Then select the most suitable 5 chromosomes as she has authority to register lecturers and students to
timetable schedules. the system.By using valid username and a password,
Crossover operations of GA create new Lecturer can log on to the system, view timetable,
individuals by combining attributes of our selected change password, view allocated resources and
individuals. As a result, crossover operator chooses mention the preferred time. Student also view
two individuals from current population (parents) timetable and change he password asrequired.
and creates a new individual (child) based on
parents’ genetic material.Here we considered no of IV. EVALUATIONANDDISCUSSION
crossover points as 2, crossover probability as 80%. 4.1 Evaluation
Using crossover ($parent2) function, made new The main goal of this evaluation was to
offspring by combining parent codes. Then, checked discuss whether the system meets the objectives
the fitness again using the calFitness(). If found a defined earlier.The significance on evaluating the
fitter chromosome than a previous selected one system was described through this system by
change it to new schedule. examines the expected output and the actual
Mutation typically works by making very output.If it satisfied our expectations, we considered
small changes at random to an individual’s that the system was behaving well.Therefore, we
genome.The mutation operator changes the value of evaluated the system with black box testing with test
some genes in an individual and helps to search data and white box testing with sample test
other parts of problem space. With regards to this casesusing specific software such as understanding
solution, our mutation probability is 3 and tool. Further, we assessed the performance and
usedmutation() to generate new chromosomes. Then robustness of the TMFIT.
again check their fitness with previous 5 best For this process, we used actual student
chromosomes using calFitness() and If found a fitter data and resources details of the faculty for the
chromosome than a previous selected one change it testing process. In that sense, the answers for the
to new one. These steps, selection, crossover, and following questions introduced through an
mutation, achieved in a 500 while loop.Then, best evaluation strategy with evaluation techniques such
five chromosomes (Timetable Schedules) as interviews, observation and questionnaires could
be used.Several interviews were conducted with
Admin of the timetable management system, some
of the lecturers and some of the undergraduate
students. System deployed in parallel way and
Admin staff gave their direct feedback about the
system functionalities. There were more than three
face-to-face interviews conducted with admin staff
and some of the lecturers and during the interview
their feedback about the system were noted down.
constructedand put them in to the flag as _bestflags System observed through sample input data for each
($chromosomeIndex). interface, their anticipated outputs and their definite
3.5 Timetable Generation output results in the evaluation phase of the TMSFIT
Timetabling System Administrator has the system. Several browser capabilities such as Google
authority of constructing timetables. Generating Chrome, Firefox, and Opera also successfully tested.
100% optimal solution from GA is not a reality,for If the system takes too much loading time, users may
an automated timetable management system. not satisfy about it. Therefore, loading time for all
www.ijmret.org Page 14
Fig.6. Generated timetable from TMSFIT
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
the interfaces had to be considered. Until it was (selection, mutation and crossover) and it is easily
hosted on a server of the campus, timetable understandable, less paper work,efficient and
component had to be given to admin staff. It was automated system, which helpful for authorities of
installed their personal computers with WAMP the IT faculty.
server. Finally, TMSFIT was further evaluated Major limitation of this TMSFIT are, the
proposed system can only generate timetables based
through a questionnaire by supplying that to
on a few hard constraints, it gives only optimal
timetable admin staff, some of the lecturers and
solutions not the best solution and it only generates
some undergraduate students. From that, we could timetables for courses and the execution time of GA
discover their satisfaction level of the system. itself is high.In future, this concept can be adapted to
4.2. Discussion fit the construction of examination timetables also.
From the admin’s point of view, overall We suggest this timetablingsystem can be used as a
system is success. Further, this TMSFIT is time framework and it will be more appropriate for
effective and user friendly. There are four batches medium scale universities.
running through the year, and we have to input all
the details of them. When the quantity system data VI. Acknowledgements
This effort was a research in utilization of timetable
increasing, timetable generation process also management system to faculty of it at University of Moratuwa,
gradually increasing and fitness value of the Sri Lanka. We acknowledge to all the senior lecturers and the
generated timetable is decreasing. As a result, staff members of the IT department at the university.Thanks also
additional manual work has to be handled. However, to great comments of the reviewers.
resource optimization phase is satisfied.
On the other hand, even, it’s set up under REFERENCES
probability theory; sometimes it supplies not the [1] E. K. Burke, D. G. Elliman, and R. Weare, “A university
timetabling system based on graph colouring and constraint
optimal but the best timetable schedulewhich manipulation,” J. Res. Comput. Educ., vol. 27, no. 1, pp. 1–
reached the fitness value as 1. Further, we could 18, 1994.
[2] D. Datta, K. Deb, and C. M. Fonseca, “Solving class
evolve the system with customer feedback such as timetabling problem of IIT Kanpur using multi-objective
adding advanced searching options, view available evolutionary algorithm,” KanGAL Rep., vol. 2006006, pp.
1–10, 2006.
resources and printing option. Therefore, after [3] J. Lee, S.-P. Ma, L. F. Lai, N. L. Hsueh, and Y.-Y.
deploying our system at the university, continues Fanjiang, “University timetabling through conceptual
modeling,” Int. J. Intell. Syst., vol. 20, no. 11, pp. 1137–
system evaluation had to be done for use it as a 1160, Nov. 2005.
framework by other universities. [4] A. Chowdhary, P. Kakde, S. Dhoke, S. Ingle, R. Rushiya,
The key research question raised in this and D. Gawande, “TIMETABLE GENERATION
SYSTEM,” Int. J. Comput. Sci. Mob. Comput., vol. 3, no. 2,
work was can an Automated Timetable Management 2014.
System solve the problem of resources optimization [5] N. Pillay, “A survey of school timetabling research,” Ann.
Oper. Res., vol. 218, no. 1, pp. 261–293, Jul. 2014.
of IT faculty? Prior literature survey suggested there [6] E. K. Burke and S. Petrovic, “Recent research directions in
isn’t a general way of solving these types of automated timetabling,” Eur. J. Oper. Res., vol. 140, no. 2,
pp. 266–280, 2002.
scheduling issues. Further, since we used only [7] B. Paechter, R. C. Rankin, and A. Cumming, “Improving a
requirements of IT facultyour research area was lecture timetabling system for university-wide use,” in
limited. Even this is workable as a International Conference on the Practice and Theory of
Automated Timetabling, 1997, pp. 156–165.
framework;application of this system to other [8] A. Colorni, M. Dorigo, and V. Maniezzo, “A genetic
faculties may slightly different. algorithm to solve the timetable problem,” Politec. Milano
Milan Italy TR, pp. 90–60, 1992.
[9] S. Hooshmand, M. Behshameh, and O. Hamidi, “A Tabu
V. CONCLUSIONAND FURTHER Search Algorithm With Efficient Diversification Strategy
for High School Timetabling Problem,” Int. J. Comput. Sci.
WORK Inf. Technol., vol. 5, no. 4, pp. 21–34, Aug. 2013.
In this paper we have introduced a model or [10] J. J. Moreira, “A system for automatic construction of Exam
a prototype for timetabling issues of middle scaled Timetable using Genetic Algorithms,” Rev. Estud.
Politécnicos Polytech. Stud. Rev., vol. 6, no. 9, 2008.
university in Sri Lanka. Even though the timetabling [11] Mp. Professor Ashoka Karunananda Bsc. PhD, Artificial
problem treated as an optimization issue, there is Intelligence, 2004.05. Tharanji Prints, Highlevel Road,
actually no fixed objective function to solve it. Nawinna, Maharagama, 2004.
[12] C.-H. Chen, T.-K. Liu, and J.-H. Chou, “A Novel Crowding
Therefore, after a proper literature survey, GA was Genetic Algorithm and Its Applications to Manufacturing
selected to construction of course timetables Robots,” IEEE Trans. Ind. Inform., vol. 10, no. 3, pp. 1705–
developed for the University of Moratuwa. This 1716, Aug. 2014.
timetabling project seeks to generate near optimal [13] “Mimosa - Scheduling Software for School and University
Timetables.” [Online]. Available:
timetables using the principles of genetic algorithm
www.ijmret.org Page 15
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 1 Issue 2 ǁ July 2016.
www.ijmret.org Page 16