International Journal of Computer Applications (0975 8887)

Volume 69 No.20, May 2013

A Multi-Objective Approach for the Project

Allocation Problem
Sameerchand Pudaruth Munish Bhugowandeen Vishika Beepur
University Of Mauritius University Of Mauritius University Of Mauritius
Port Louis, Quatre Bornes, Vacoas,
Mauritius Mauritius Mauritius

ABSTRACT
In this paper, a novel system is presented for the allocation of The project work is assigned at the beginning of Semester 1,
final year projects for the Computer Science and Engineering that is, at the start of August and must be submitted by the end
Department at the University of Mauritius. Earlier works had of March in the following year. Each student or team of two
concentrated only on the allocation of projects to students. or three students is expected to spend about ten hours per
The system not only performs project allocation but it also week over twenty-six weeks on their projects. The final year
allows academics to rate projects, examiners to bid for project provides the student the opportunity to develop their
projects they wish to examine, students to propose their own problem solving, analytical skills and evaluation and
projects, students to submit project deliverables, supervisors programming skills. A typical project format would require
to follow projects more closely and allows projects the student to see how some aspect of theory, software or
coordinators to have a heuristic view of the whole system. hardware that they are familiar with and that can be enhanced
The system captures the preferences of examiners as well as or developed into a product or a hardware/software tool.
students and allocates projects to them in order to maximise
the number of students who gets their first choice in their
2. BACKGROUND STUDY
preference list and to keep the load of supervisors and The problem of project allocation crops mainly in the
examiners within a reasonable range. The percentage of Computer Science and Engineering department in the
students who obtained their first choice is 82% on 30 projects University. It has a student population of about seven
proposed by 15 supervisors for 11 teams. The simulation hundred. The CSE department consists of three undergraduate
results demonstrate that this new system will allow deadlines programmes which are the BSc (Hons) Information System,
for all the different project phases to be met. the BSc (Hons) Computer Science and the BSc (Hons)
Computer Application. The lecturers who teach in these three
Keywords programmes are the same. Each year the CSE department
Project Allocation, Examiner Allocation, teams, supervisors, experiences a problem while allocating projects for the final
resources, projects, examiner bidding, project rating year students. These problems are mainly due to the fact that
year after year, the number of students keeps increasing and
1. INTRODUCTION this leads to an increase in the complexity in the allocation of
the projects. There are also constraints such as supervisors
The Computer Science and Engineering (CSE) Department of workload which plays an important role in the allocation of
the University of Mauritius is composed of 36 academic staffs projects. All supervisors should have more or less the same
supported by 9 technical staffs. It is one of the youngest amount of projects to supervise. The lecturers need to propose
departments of the Faculty of Engineering. It is devoted a certain number of project titles for all the three programmes.
towards quality teaching, research in advanced areas of
computer science and consultancy work. The department is Together with the allocation of projects and supervisors,
equipped with a number of laboratories, where students have another important allocation is that of the examiners.
facilities, in terms of hardware, software, network Examiners will be the person in charge to assess the final year
infrastructure and internet access to be able to carry out their projects. One examiner is currently assigned for each
practical work and assignments [1]. undergraduate project. This is done in a transparent way
according to their area of expertise and experience in their
For every undergraduate student of the University of respective fields.
Mauritius, there is a final year project (FYP) which needs to
be taken. The project carries nine credits and a report must be The allocation of projects is part of a degree course for most
submitted for marking. Furthermore, some faculties have universities. However, before the allocation of projects to
viva-voces while others have oral presentation. students is done, there is a series of steps which need to be
performed. Many constraints need to be taken into
Resource allocation is the distribution of the available consideration to achieve a fair allocation. The allocation of
resources in an economically manner. The allocation of project causes many problems as same constraints may not be
project is like a subset of the resource allocation problem. satisfied. The aim is to achieve a better system which will
The allocation of projects should be done fairly for both the solve all these problems or at least reduce them
lecturers and students [2]. The resources are the projects, significantly.In the previous years there were only two
students and academics. The same academics act as both programmes in the CSE department which were the BSc
supervisors and examiners but for different projects. (Hons) Information System and BSc (Hons) Computer
Science and Engineering. During these years, the project titles
proposed were for both Information System and Computer

26
 International Journal of Computer Applications (0975 8887)
Volume 69 No.20, May 2013

Science and Engineering students but this was later found to solution, there are different cycles which are carried out, for
be a problem since the syllabus of the programmes were the allocation of projects starting from the first choice for
different. This was unfair to many students as students from everyone then with the remaining student their second choice
the BSc Information Systems stream were working on is taken into consideration. This decreases the number of
projects involving largely on Computer Science disciplines unassigned projects. If there is still student unallocated to
but they were never thought such subjects. The same would project a second is done, and the student will make choices
be said for BSc Computer Science students who had to work again. This process could have been easier if the allocation
on the development information systems. This created a instead of making different rounds were done for one student
problem, so the allocation of projects is now done separately at a time that is, if the student first allocation were not done
for each programme. then his second to tenth choice is checked and allocation is
done.
The distribution of the projects is made such that each lecturer
supervises a certain number of projects, whether the projects Moreover, in [5] the system matches students having
are done in groups of two or individually. Ideally, the preferences over projects along with the lecturers having
allocation of projects should be done in one round. However, preferences over students. Students are first matched to the
in some cases there are some students who are not allocated best project that he could obtain. The algorithm is applied
any projects after the first round, so a second round is again so that the lecturers could get the best group of students
performed. This can happen when students do not submit their to supervise. In this system, students having low academic
project selection form on time, they have not yet formed a grading can be penalised. To solve this problem, the SPA-P
team or when students are not allocated any of the projects in [5] algorithm was implemented. This handles cases where two
the list as all of them have been allocated to other students. A or more lecturers would like to choose the same group of
third round is sometimes necessary as well. students and cases where students have not been selected by
any lecturer.
Projects examiners also need to be assigned. In previous
years, examiners were assigned one week before the final In [6], before the start of semester one, each staff member
presentation of the project. This created problem like (project supervisor) is requested to submit up to four project
examiners found that projects given to students did not had an titles together with a small description. They are also expected
adequate level of complexity suitable for a final year project to give any other piece of information that will help the
or that their field of expertise did not match the project they student to understand the project well. Each project is given a
are given to examine. Also they did not have sufficient time to unique project number and is hosted online (WebCT) for the
understand the whole project. This penalised students as they students to access it. In the first week, students are allowed to
had no other opportunity to improve their work. For the browse through the list of projects and discuss them with the
current academic year (2012-2013), there has been a major respective supervisors. However, no project will be allocated
change in the system. Examiners are now allocated to projects until the start of the second week. Project allocation starts in
at the very beginning of the project management phase. This week two and the process is conducted via email. A project is
will help to eradicate the problem faced in the previous years. allocated to a student if both parties, that is, the student and
A poster presentation was also held in mid-January to allow the supervisor agreed and confirmed on the same project
all academics, especially project examiners, to have a look at number. The concept of first come first served is applied.
the projects and to suggest improvements. However, as in With this system there is still the problem of popular titles
previous years, students do not know who their examiners are. where large amount of students are attracted to a small
By doing so, it is expected that the whole system will now be number of projects.
fairer to everyone.
In [7], supervisors have to submit a list of their proposals in a
3. LITERATURE REVIEW structured format to the secretarial staffs. A document with all
project titles is then provided to students to make their
[3] presents two solutions for the project allocation problem. choices. Later in the year, students are allowed to meet the
In the first model, every student is allocated one project at supervisors and discuss about the projects. During this
random so that each supervisor has at least one project to meeting, the supervisor elaborates on the skills required for
supervise. In the second model, students choosing the same the particular project. He can then advise the student whether
title will form groups so that they can work on the same he can do this project or not. However, the final decision of
project. The first solution is hardly a solution as students selecting the project remains entirely on the student. Since it
cannot choose on which projects they wish to work. For the is almost impractical for a supervisor to meet all the students
second one, the students who form groups may not know each twice, allocation is normally done on a first come, first
other well and may have some communication difficulties in served basis even though the student may not have all the
the beginning. There are also cases where more than two appropriate skills for the project.
students select the same project. This solution does not cater
for a fair distribution of project, supervision workload across A technique known as Goal Programming formulation [8]
supervisors. attempts to allocate a maximum number of projects taking
into consideration both students and staff preferences. The
model in has three hierarchical goals. The first and the most
important goal is to allocate the maximum number of projects.
In AssignProj [4], initially, lecturers were asked to submit up
Secondly, to maximise the total satisfaction of the students
to three projects which are will be evaluated based on the
and thirdly, it minimises the dependency of Grade Point
level of investigation and are then modified accordingly. After
Average (GPA) for the allocation of the projects. A fuzzy
this process, students are emailed these projects for them to
algorithm was proposed in [9] to solve the project allocation
make their selections. They will have 10 selections to make
problem. The students have to submit a list of eight projects
and they will be assigned to their first choice unless there is
in order of preference. Their Cumulative Average Point
contention where they will be allocated to any random project
(CAP) is used to classify the students from zero to five. The
found in their respective selections. With the new proposed
algorithm uses a probabilistic approach (Pareto-Optimal

27
 International Journal of Computer Applications (0975 8887)
Volume 69 No.20, May 2013

Method) to assign projects to appropriate students taking into seniority list is used to resolve contention instead of CPA. The
consideration its priority, the supervisors workload and a complete algorithm is shown in Fig. 1.
balanced spread of good students across supervisors. Students
who do not obtain any projects are allocated a random project 5. EVALUATION OF RESULTS
from the list of unallocated projects.
Using thirty projects for eleven teams and fifteen supervisors,
The Bioscience departments, within the UK Higher Education there are 82% of first choices which are allocated to teams.
(HE) [10], conducted a survey to identify their method of Only students from the BSc (Hons) Information System
allocation. It was found that 69% of the departments allocate programme were considered. Each team has a priority list in
project based on the list of preferences submitted to students. which the projects and project choices are saved. Number one
64% allocates projects based on direct discussion of the has the highest priority while number ten has the lowest.
student with academic staff, 50% uses student performance to
allocate projects if there is contention, 43% ask students to
indicate their own area of interest and the supervisor with Table 2. Percentage of First Choice in Manual System
whom they want to work with, 14% also ask their students to
identify the fields they do not want to work in and finally, BSc BSc BSc
4.3% allocate projects randomly. Another interesting study Course Information Computer Computer
was done in [11] to compare, contrast and evaluate the System Science Application
different procedures that are used to allocate projects.
First
9 12 8
There is considerable work in the literature, which proposes Choice
solutions for the project allocation problem. However, none of Manual Non-First
them have created a benchmark which allows comparisons 7 15 9
System Choice
between the outputs from the different systems. Thus, in this
paper, a new framework has also been proposed to allow % of First
56 44 47
researchers and other institutions to compare their work with Choice
each other. Table 1 shows the features of existing works.
Table 1. Comparative Analysis of existing techniques Table 2 shows the percentage of students who have been
allocated their first choice in each of the three programmes.
Lecturer Student How This results obtained are based on a sample of students only.
Student
Second Nevertheless, it is seen that the percentage of students
Project Selects Proposed contention
Round obtaining their first choice is considerably less than what was
Selection
Student Project is resolved? obtained in the proposed system.

[3 Yes No No No Group students

]
with contention
[4 Yes No Yes Yes Assessments
] 6, 7, 8,
marks
[5 Yes Yes No No Not available
] 1, 2, 3 9, 10
[6 Yes Yes No No FCFS 4, 5
] Very Good Very Poor
[7 Yes No No No FCFS
]
Yes
[8 No No No No GPA
]
No
[9 Yes No No No Random
Figure 1. Satisfaction Framework
Yes allocation
This range is based upon the level of satisfaction that a team
4. DESIGN AND IMPLEMENTATION receives after obtaining the result of the project allocation
process. A performance metric which will give us an overall
The project allocation algorithm is based on the principles of assessment of the efficacy of the allocation made by the
the Marriage algorithm [5] which helps to allocate projects to system has been derived. This metric classifies the first to
teams. Students forms team and register themselves in order third projects choices as a very good allocation, the fourth to
to perform project selection. Each team needs to select ten fifth choices as satisfactory and the others as very poor
projects from their respective programme. Examiners are allocation. A ratio known as the Average Priority of Choice
only allocated projects for which they have shown interest. Allocated (APCA) is computed to assess the whole system.
Firstly, all the first choices of the different teams are retrieved. The APCA is obtaining by adding the priority of each
The project is allocated to the team unless there is contention allocated project from the initial allocation list divided by the
in which case the team with the highest CPA is allocated to he total number of projects allocated. In this simulation, nine
the project. Next, the second choices for all students who have teams got their first choices, one team got its second choice
not yet been allocated a project are considered. This process is and one team got its third choice. Thus, the APCA produces a
repeated until all students have been allocated a project. Each ratio of 1.27, which when compared with the classification
project can only be allocated to only one team. Sometimes, it above, implies a very good allocation.
becomes necessary to perform a second round if ever there are
some students who have not obtained any project. Examiner
allocation is similar to project allocation except that a

28
 International Journal of Computer Applications (0975 8887)
Volume 69 No.20, May 2013

START No

Is CL Retrieve team id, project

Retrieve all projects = 1? code and allocate project
Yes to team

Get choices, C from 1 to

10 No
Remove project selection Is CL
in CL > 1?
For each project, retrieve
all project code, PC No
Yes

Is position Yes
No of project
= -1?
Is Project Code Retrieve team
= PC and id, high CPA
choice = C in
project
selection?
For each project selection in
AA, get position of project in Allocate project to
CL END team with highest
CPA
Yes

Retrieve the project selection Retrieve project selection from

into Contention List, CL CL to allocated array list, AA

Yes

If contented
No
Retrieve the Project team = ATI or
Allocation, PA according to contented
the PC into Allocated List, project =APC?
AL

For each project allocation Retrieve team id, ATI and

in AP project code, APC

Figure 2. Allocation of Projects

29
 International Journal of Computer Applications (0975 8887)
Volume 69 No.20, May 2013

6. CONCLUSION [4] Teo C. and Ho D., "A systematic approach to the

implementation of final year project in an electrical
The simulation results demonstrate that the multi-objective engineering undergraduate course," IEEE Trans. Educ.,
resource allocation algorithm is able to provide solutions to Vol.41, No.01, 1998.
many of the problems inherent in allocation of resources. The
solution not only caters for the allocation of projects to [5] Manlove D. F. and OMally G.: Student-Project
students but it also has many innovative features which have Allocation with Preferences over Projects, Journal of
never been considered before in previous works. Indeed, the Discrete Algorithms, Vol.6, No.4, 2008.
system also handles the allocation of examiners to projects [6] Gallagher Alison M., McKerr George, Gill Chris I.,
based solely on their interest to supervise a project or not. A Streamlining allocation and assessment of traditional
performance metric has also been devised to allow different final year research projects across multiple
allocation systems to be compared and hence evaluated. This undergraduate degree programmes, Centre for
metric shows that the algorithm does extremely well in Bioscience, The HE Academy, 2008.
allocating projects. With slight modifications, the system can
also be adapted to manage assignments and mini-projects as [7] Dimitar Kazakov, Coordination of Student Project
well. In the future, different algorithms will be implemented Allocation, October 31, 2001.
to do the allocation and then they will be compared it with the [8] Pan L., Chu S. C. K., Han, G., Huang, J.Z., Multi-
current one. criteria Student Project Allocation: A Case Study of Goal
Programming Formulation with DSS Implementation,
7. REFERENCES The Eighth International Symposium on Operations
[1] University of Mauritius. 2013. University of Mauritius. Research and Its Applications (ISORA09), Zhangjiajie,
ONLINE] Available at: China, September 20-22, 2009, ORSC & APORC.
http://www.uom.ac.mu/Faculties/foe/CSE/overview.htm.
[9] Rachmawati, L. and Srinivasan D., A Hybrid Fuzzy
[Accessed 27 March 2013].
Evolutionary Algorithm for A Multi-Objective Resource
[2] Wikipedia, the free encyclopedia. 2013. Resource Allocation Problem, In Proceedings of the Fifth
allocation - Wikipedia, the free encyclopedia. [ONLINE] International Conference on Hybrid Intelligent Systems,
Available at: IEEE, 2005.
http://en.wikipedia.org/wiki/Resource_allocation. [10] Richard J. Cowie, A Snapshot of Final Year Project
Practice in UK Bioscience Departments, Centre for
[Accessed 27 March 2013]. Bioscience, The HE Academy.
[3] Anwar A. A. and Bahaj A. S., Student Project [11] Janice Harland, Sarah Pitt, Venetia Saunders, Factors
Allocation Using Integer Programming, IEEE affecting student choice of the undergraduate research
Transactions on Education, Vol.46, No.3, 2003. project: staff and student perceptions, BEE-j Volume 5,
May 2005.

30