Download - Zucas Final
-
8/17/2019 Zucas Final
1/69
Project Topic:
ZIMBABWE UNIVERSITY CENTRALISED ADMISSION
SERVICE
By
KURASA TOREVEI
St!"e#t N!$%er: &'&&()*M
De+ree: BTec, -o#./ "e+ree i# IT -B0oc1 Re0e2.e/
S!per3i.or: Mr45otor2 O
-
8/17/2019 Zucas Final
2/69
This Project is submitted to the Department of Information Technology , Harare Institute of Technology, in
partial fulfillment of the Bachelor of Technology Honors Degree in Information Technology
-
8/17/2019 Zucas Final
3/69
S!per3i.or: Mr45otor2 O
Si+#2t!re:66666666666666
-
8/17/2019 Zucas Final
4/69
ACKNOWLE5EMENT
Firstly and foremost, I would like to thank the Almighty GOD for guiding me through my
studies and the subsequent work on this project !y special thanks e"tends my wife Dorothy
and son #ore for the sacrificed joy during my study time
Of course, like any author, I am indebted always to those people that do their best to impro$e
on my best #hanks especially to all who ha$e checked on my work and corrected some te"t
in this documentation and this was the work of my super$isor !r Gotora O who ga$e me
resounding guidance and leadership during the de$elopment of this project
#o the department of computer science and information %ciences and #echnology students
and staff, I really thank you for your unwa$ering support
#hank you, and be &ell and 'appy
-
8/17/2019 Zucas Final
5/69
ABSTRACT
#he adoption of information and communication technology has seen the automation of most
of the uni$ersity processes which has since proliferated in some uni$ersities, with the
admission process being one of the processes being automated #he necessity of fully
e"ploiting the use of information technology has enlightened the need to design a centralised
uni$ersity admissions system that addresses the specific requirements for the (imbabwean
uni$ersities #his thesis del$es into the design of an effecti$e centralised admissions system
for undergraduates, which can sol$e some problems which are currently being encountered
with the decentralised admission system #he system model was e$aluated using a
combination of software performance tests and perceptions from the system users
-
8/17/2019 Zucas Final
6/69
#able of )ontents
)'A*#+ O-+. I-#OD/)#IO-0000000000000000000000001
12 Introduction00000000000000000000000000000001
11 3ackground of study0000000000000000000000000001
14 *roblem statement0000000000000000000000000000415 %tatement of purpose0000000000000000000000000004
16 esearch objecti$es00000000000000000000000000004
17 esearch questions00000000000000000000000000005
18 9ustification00000000000000000000000000000005
1: Assumptions0000000000000000000000000000006
1; s +nd56
557 %ystem %ecurity 58
558 %election Algorithm 5:
55: *ost %election 5;
55; %ystem testing and e$aluation 5;
56 *opulation and %ampling 5;
57 Data Analysis *rocedures 5=
)'A*#+ FO/. DA#A *+%+-#A#IO- A-D A-A
-
8/17/2019 Zucas Final
7/69
72 Introduction82
71 Aims and objecti$es realisation 82
74 Future &ork82
-
8/17/2019 Zucas Final
8/69
#able of Figures
Figure 1 . )entralised /ni$ersity Admissions %ystem use case diagram 44
Figure 4 )entralised /ni$ersity Admissions %ystem acti$ity diagram 46
Figure 5 )entralised Admissions %ystem process flow diagram 48
Figure 6 )entralised /ni$ersity Admissions %ystem entity relationship diagram 4;
Figure 7 /ser interface design process 52
Figure 8 Account creation page for applicants 51
Figure : +ntering qualifications for an applicant 54
Figure ; making an application on the system 55
Figure = %etting degree selection criteria56
Figure 12 !anaging uni$ersities in the system 58
Figure 11 Adding a new faculty 58
Figure 14 *rior e"perience with internet 68
Figure 15 the centralised admission system is reasonably easy to use 6;
Figure 16 )arrying out the application process was easy 6=
Figure 17 the system is unnecessarily comple" 72
Figure 18 O$erall system rating +rrorB 3ookmark not defined
Figure 1: Internet connecti$ity 74
Figure 1; Ability to access platform 75
Figure 1= +ase of use of the centralised system 76
Figure 42 O$erall system rating 78
Figure 41 eduction in redundancies 7;
Figure 44 human intensity reduction 7=
-
8/17/2019 Zucas Final
9/69
CAPTER ONE: INTRODUCTION
&4* INTRODUCTION
#he fast de$elopment of information and communication technologies CI)# and the
e"pansion of the internet ha$e changed different industry structures around the world #oday
organiEations of all types around the globe are utiliEing I)#, not only for cutting costs and
impro$ing efficiency, but also for better ser$ice pro$ision 'owe$er, in (imbabwe, the use of
I)# for uni$ersity undergraduate admissions is not being fully e"ploited
/ni$ersity undergraduate candidate selection is the process through which applicants are
accepted for uni$ersity entry to undertake undergraduate degree programs In (imbabwe,
most prospecti$e uni$ersity undergraduate students apply for admission in differentuni$ersities upon the completion of ad$anced le$el studies Admission in any of the country>s
state uni$ersity differs due to the selection criteria used in the selection process by each of
these uni$ersities
ey aspects of the current selection system that will be of note would be the ability of the
proposed centralised system to reduce the number of simultaneous multiple offers to one
particular applicant by differing uni$ersities, speed of placement and consistency of the
centralised system to gi$e consistent results o$er $arying data that should not ha$e effect on
its selection mechanism
A presumption is made that a softwaredri$en system is most likely to be more effecti$e in
most parts than the present manual, decentralised system #hus, the proposition is that this
research be summarily composed of an e$aluati$e step of assessing the practicality of design
of the system ie the design of an alternati$e centralised candidate selection system
&4& BACK5ROUND O9 STUDY
As an attempt to address the shortcomings of the con$entional undergraduate candidate
selection system in uni$ersities, a centralised system which can be used by all applicants to
apply to any of the state uni$ersities is going to be de$eloped
#o date, the candidate selection process is decentrally carried out at each of the uni$ersities,
which has resulted in redundancy of the selection process Applicants are being allocated
places in more than one uni$ersity in utmost cases yet each applicant can take up a place in
-
8/17/2019 Zucas Final
10/69
only one uni$ersity lea$ing the other allocated places $acant In most of the uni$ersities, the
candidate selection process is still being done manually which consumes a lot of time
#he acceptance of post As selection criteria for the degree program they will be applying for
/ndergraduate entry selection criteria and standards $ary from one uni$ersity to another and
these will be ascertained in the uni$ersitiesH admissions policies
&4' PROBLEM STATEMENT
#he lack of a fast, centrally decisi$e managed process has yielded undesirable results in theundergraduate entry system across state uni$ersities in (imbabwe #he current state of affairs
has led to predominant redundant offers, time and resource consuming candidate selection
process for undergraduate entries in each of the state uni$ersities %imultaneous multiple
offers to one particular applicant by differing uni$ersities ha$e to be eliminated Applicants
ha$e to be informed in time, of the a$ailability of their enrolment places so as to reduce
incon$eniences for both the applicants and the admission staff in these uni$ersities Also, the
current way of selecting candidates for uni$ersity entry tends to be not as accurate, as it is
prone to human errors and bias since it is done manually 'owe$er, adopting e"isting
solutions such as the deferred acceptance algorithm does not quite suffice the problem as they
tend to ha$e their own weaknesses
15 %#A#+!+-# OF */*O%+
#o de$elop a centralised uni$ersity undergraduate candidate selection system that is capable of
allocating uni$ersity places to applicants in a fast, accurate and nonredundant manner
16 O39+)#I+%
• #o design and implement a centraliEed uni$ersity undergraduate candidate selection system
model, using the stable matching concept
• #o e$aluate the usability and effecti$eness of the resultant system
• #o de$elop a fully functional design of the centralised uni$ersity admissions system
17 9/%#IFI)A#IO-
#he con$entional candidate selection process in state uni$ersities has shown some weaknesses and the
researcher has therefore seen it feasible to research on how best the candidate selection process can be
-
8/17/2019 Zucas Final
11/69
centraliEed for undergraduate entry across state uni$ersities, addressing redundancies posed by the
presentday selection process 3y centralising the candidate selection process, the problem of multiple
simultaneous offers to one applicant by different uni$ersities can be eliminated #he centralised
system will also benefit each of the uni$ersities through a reduction in resource wastages which arise
due to the simultaneous multiple offers )entralising the candidate selection process would mean that
the number of students who will turn down admission offers because they would ha$e been offered
admission by another uni$ersity will be greatly reduced In such a way, resource wastages by the
uni$ersities will be minimal as a greater number of students offered admission will accept, thereby
aligning with the uni$ersities> admission plans
+fficiency in the selection process will also be increased through the proposed system 3ecause the
selection process is currently being done manually, the centralised system will reduce the time taken
for candidate selection, and this in turn would mean that candidates will be notified in time of the
a$ailability of their enrolment places Also, de$eloping the proposed system results in a candidate
selection process that will be free from human errors and bias hence the candidate selection process
will be more accurate and in a way, fair
For the technical implementation feasibility test of the system, we base this on the assumption that the
process of the acceptance of the system does not affect change to the suggested e"ecution of the
proposed system
18 s academic qualifications
through the national e"amination results pro$iders which are (I!%+) and )ambridge Acquiring the
actual results databases from these e"amination boards for the purpose of this research is likely to be
not $iable due to the confidentiality policies of these boards
1: %)O*+
#he model to be de$eloped shall only cater for the admission of post Ale$el students for
undergraduate programmes #his implies that applicants with other forms of qualifications will not be
catered for in the system due to time reasons Although the proposed system will ha$e the capability
to accommodate all state uni$ersities, only three of the state uni$ersities shall be used namely 3indura
/ni$ersity of %cience +ducation, where the researcher is based, /ni$ersity of (imbabwe and the
-ational /ni$ersity of %cience and #echnology
-
8/17/2019 Zucas Final
12/69
CAPTER TWO: LITERATURE REVIEW
'4* INTRODUCTION
e$iewing related literature will assist the researcher to unco$er what other researchers ha$e done on
this subject matter #his chapter re$iews rele$ant research made on the design of centralised
admissions systems, particularly the algorithms employed )ollecti$e ideas and $iews of different
authors are contained in this chapter #he ser$es to answer the following questions what are the
current algorithms that are being used in the designing of centralised admission systems and what
gaps need to be filled to impro$e the e"isting algorithmsJ
#he college admissions problem is an e"ample of a two sided matching market which has been widely
studied by economists and game theorists CAbdulkadiroglu A 4227 #his is whereby a college only
accepts a specific number of students Cquotas in each academic year due to the impossibility of the
college to accept all students who applied for it because of limited resources %o, e$ery student cannot
get into their top choices and on the other hand, a student also can only accept offer of admission from
only one college, thus it is not guaranteed that all students whom a college has made offers of
admission will accept the offers
Issues in$ol$ing who gets which job, who gets which school places and the admission of students to
colleges are among the matching situations that ha$e gained attention in the last decades In a two
sided matching market, agents belong to one of two disjoint sets, for instance colleges and students,and each agent ie college and student, has preferences o$er the other side of the market ie students
and colleges, and the prospect of being unmatched *erhaps the most famous matching mechanisms
are the 3oston mechanism and the Gale%hapely algorithm #hese two differ in the criteria applied for
admission under each of the mechanisms !any countries around the world are using the centraliEed
admission system, for e"ample German, %weden and 'ungary, which are all based on either the
3oston mechanism or the Gale%hapely algorithm
In this research, the effecti$eness of the stable matching algorithms in pre$ious matching problems is
analysed and a further study is made to attempt to analyse the root causes of the weaknesses and
limitations associated with these particular solutions in sol$ing the stable marriage problem in
comple" markets where the matching is not simply one entity in set A being matched to only one
entity in set 3 Csimple markets as initially e"postulated by Gale and %hapely C1=84 A critical
analysis is then made to formulate sufficient facts on which to impro$e the e"isting algorithm for use
in the case of the centralised uni$ersity admissions system as a comple" market
'4& OVERVIEW O9 CENTRALISED ADMISSIONS SYSTEMS
-
8/17/2019 Zucas Final
13/69
#he centralisation of the uni$ersity admission process has been appreciated across the world In
'ungary, student admissions for both secondary and higher education are organised by centralised
matching schemes #he programs for both the secondary schools and higher education are based on
the original model and algorithm of Gale and %hapley C3iro et al 4212
#he centralised German uni$ersity admissions system implements the primiti$e prejudicial
mechanism to be e"plained later, the K3oston mechanismH CAtila Abdulkadiroglu L #ayfun %onmeE
4225, and the Deferred Acceptance Algorithm by Gale and %hapley Although for the majority of the
subjects, the uni$ersities decentrally select the students themsel$es, there is a centralised matching
scheme administered by the clearing house that is used to allocate places for medicine and three
specialitiesCdentistry, pharmacy and $eterinary medicine as these fields of study are most prone to
o$er demand in the country C3raun, ubler and Dwenger 4212
'4' ORI5INS O9 TE STABLE MATCIN5 PROBLEM
#he stable matching Cmarriage problem originated, in part, in 1=84 when Da$id Gale and
-
8/17/2019 Zucas Final
14/69
Abdulkadiroglu and %onmeE find $arious shortcomings of the 3oston mechanism. the mechanism is
not stable, that is, it can cause an unfair outcome where a student is not admitted to a school they like
while a student with a lower priority than them is admitted to that school the 3oston mechanism is
not strategyproof, that is, it gi$es students a strong incenti$e to misrepresent their preferences by
impro$ing the ranking of schools for which they ha$e high priority #he authors also highlighted the
mechanism>s ease of manipulation #he authors further point out that the deferred acceptance
mechanism by Gale and %hapely sol$es both problems because it is both stable and strategyproof
Gale and %hapley focused on a common problem faced by colleges based on their usual admissions
procedure, namely, how to admit the ideal number of bestqualified applicants based on a specific
quota without knowing precisely how many admitted applicants will accept CGale and %hapley 1=84
For e"ample, the authors note the college may not know whether the applicant has applied elsewhere
and, if so, whether the other colleges will admit them For the stable matching problem, Gale and
%hapley analyEed matching at an abstract general le$el #hey used marriage as one of their illustrati$e
e"amples how should ten men and ten women be matched while respecting their indi$idual
preferencesJ #his in$ol$ed a set of men and a set of women, each whom ha$e ranked all the members
of the other set in a strict order of preference #he main challenge in$ol$ed designing a simple
mechanism that would lead to a stable matching, where no couples would break up and form new
couples that would make them better off #he authors used their solution to this problem as a basis for
sol$ing the e"tended problem where one of the sets consists of college applicants, and the other
consists of colleges, each of which has a quota of places to fill
'owe$er, this was not the only origin of the stable matching problem It turns out that for a decade
before the work of Gale and %hapley, unbeknownst to them, the -ational esident !atch *rogram
had been using the $ery similar procedure Cleinburg and #ardos 4228 #he program began in 1=74
in response to dissatisfaction with the process and results of matching applicants to residency
programs $ia the decentralised, competiti$e market #he -!* administers the matching of
graduating medical students to hospital residency positions through a centralised ser$ice Coth and
%otomayor 1==2 From shortly after the first residency programs were formally introduced, thehiring process was characterised by intense competition amongst hospitals for interns A publication
in 1=84 by Gale and %hapley noted that there always e"ists a stable solution when colleges are
matching with students, but that it is possible to fa$our colleges as a group o$er applicants as a group
/p to the le$el of simple markets, matching in which only onetoone matching is the only possibility
Csimple markets, and no couple Cor more of accepting agents would rather prefer a match in which
they outright want combined acceptance by one proposing agent Ccomple" markets the deferred
acceptance algorithm has been pro$en to work and always produce stable matching CGale and %hapley
1=84 Coth A 422; #he authors further pro$e that this algorithm fails and will not always produce
-
8/17/2019 Zucas Final
15/69
a stable matching in a case where couples e"ist #hey faced this problem where some of the accepting
medical students would be married couples who would rather prefer a matching to a hospital where
they both are accepted
'4) TE DE9ERRED ACCEPTANCE AL5ORITM
In the deference algorithm C!iralles A 422;, students submit their ranking lists that correspond their
preferences o$er the colleges they are applying for #he students then recei$e scores at each college
they applied for according to their final grades at secondary school, and entrance e"ams #hese scores
will be a real number between 2 and 1 assigned by an e$en lottery #he assignment is computed in
se$eral rounds where in the first round, the mechanism assigns each student to her first ranked school,
in increasing order of the lottery numbers, until either school capacity is reached or no more students
rank the school as their first choice In each of the remaining rounds, each student applies to their
most preferred school amongst the ones that would not ha$e rejected them Again, the algorithm
compares all students applying to the school and reassigns students in their increasing order of lottery
numbers until full capacity is reached or until no students remain
#he Gale%hapley algorithm has been found to be stable in that no student loses a seat to a lower
priority student and recei$es a less preferred assignment and also strategy proof for students, that is, it
is a dominant strategy for students to state their true preferences ariants of the college proposing
deferred acceptance C)DA algorithms ha$e been used to allocate medical students to their
professional position in the /% Coth 1=;6 and $ariants of the student proposing deferred acceptance
algorithm C%DA ha$e been and are still being used to assign students to public schools in 3oston and
-ew ?ork CAbdulkaroglu et al 4227
'4 TE COLLE5E PROPOSIN5 DE9ERRED ACCEPTANCE AL5ORITM
In the college proposing deferred acceptance algorithm, each college proposes to its top acceptable
students +ach student rejects any unacceptable proposals and, if more than one acceptable proposal is
recei$ed, the student holds the most preferred and rejects the rest Any college which would ha$e been
pre$iously rejected by any student proposes its most preferred acceptable students who would not
ha$e rejected it +ach student holds their mostpreferred acceptable offer to date and rejects the rest
#he algorithm terminates when there are no more rejections +ach student is matched with the college
they would be holding in the last step
According to C3raun, ubler and Dwenger 4212, in the German>s centralised uni$ersity admissions
procedure the 3oston mechanism is used to firstly allocate up to 62 percent of the total capacity of
each uni$ersity and all the other remaining places are assigned using the college proposing deferred
acceptance algorithm studied the German uni$ersity admissions system from an empirical perspecti$e
-
8/17/2019 Zucas Final
16/69
and found considerable support for the hypothesis that applicants try to manipulate the centralised
admissions procedure Applicants ha$e strong incenti$es to manipulate the German admissions
procedure by submitting a ranking of uni$ersities that do not correspond to their true preferences
#he way in which conflicts are resol$ed in the 3oston mechanism and the Deference Algorithm isdifferent #he deference algorithm makes truthful ranking a dominant strategy and resol$es any
conflicts purely by random lotteries regardless of the underlying cardinal utilities #herefore the
outcome of the deference algorithm is completely insensiti$e to cardinal preferences In contrast, the
3oston allows parts of participants to influence how ties are broken, so it has the potential to resol$e
conflicts based on students> cardinal utilities which may be useful for efficient resolution of
conflicting interests )onflicts arise due to tension in stabilising equity in assignments, that is, the
e"tent to which student priorities can be accommodated, student welfare and immunity to strategic
action CA + oth 1=;4
#he 3oston mechanism may e"pose strategically nai$e participants as pro$ed by Abdulkaroglu et al
that some players may ha$e beha$ed nai$ely and suffered as a consequence under the 3oston
mechanism %chools that are o$erdemanded tend to enable $aluable traders, as there are many
students who would be willing to e"change their assigned school for an o$erdemanded one #hus
misreporting would make o$erdemanded schools to be e$en more o$erdemanded #he increase in
demand amplifies trade opportunities, and imposes a negati$e e"ternality on the other students
%tudents may e$en report a high preference for a school they do not find acceptable as long as it
enables desirable trades &hen trade is not certain students may still attempt it, but when they fail to
trade they can end up with an unacceptable school, making the final matching unstable and inefficient
A theoretical justification to 3oston mechanism>s weaknesses was pro$ided by *athak and %onmeE in
422;, who argue that strategically sophisticated participants e"ploit nai$e ones in the 3oston
mechanism, to such an e"tent that the former effecti$ely enjoys a higher priority o$er the latter at
e$ery school e"pect for the latter>s most preferred
'4( TE PROBLEMS O9 TE DE9ERRED ACCEPTANCE AL5ORITM
As confirmed by Coth A 422;, the problem of couples, e$en after attempts to make the deferred
algorithm e"ists still In our case, this is concerned with a typical situation where a uni$ersity is
considering a set of applicants which it can admit only a quota a pair or more of these ha$ing a gi$en
KconditionH to only be accepted, e$en if one outright meets the criteria and prefers when another
student which that particular also prefers to be accepted is accepted at that same uni$ersity 'a$ing
e$aluated their qualifications, the admissions office must decide which ones to admit #he procedure
of offering admissions only to the best qualified applicants will not generally be satisfactory, for it
cannot be assumed that all those offered admission will accept Accordingly, in order for a college to
recei$e a particular number of acceptances, it will generally ha$e to offer to admit more than a certain
-
8/17/2019 Zucas Final
17/69
number of applicants CGale and %hapley 1=84 According to the authors, the problem of determining
how many and which ones to admit requires some rather in$ol$ed guesswork since it may not be
known whether a gi$en applicant has also applied elsewhere if this is known it may not be known,
and also which of the other colleges will offer to admit him
#he deferred acceptance algorithm by Gale and %hapley sol$es the stable marriage problem through a
series of MproposalsN made by the MmenN, resulting in pro$isional MengagementsN that the women are
free to break later if they recei$e a better offer In practise, because the deferred acceptance algorithm
assumes strict preferences, almost all matching schemes either require all participants to rank their
choices in strict order of preference, or use some form of randomisation to break any ties in the
preference list #his may result in arbitrary decisions, either on the part of participants or scheme
administrators, who produce a strictly ordered list by discriminating unnecessarily or artificially
between applicants
#hrough the deferred acceptance algorithm, Gale and %hapley did pro$e that the set of stable
outcomes is nonempty for all preferences of the kind they considered and also that when the
preferences are strict, there always e"ists a stable matching Coth A 422;&ith the student proposing
deferred acceptance algorithm, the mechanism specifies how to order equi$alent students from the
point of $iew of schools with limited space For instance, one can assign each student a distinct
number, break ties in school preferences according to those assigned numbers, then apply the deferred
acceptance algorithm to the strict preferences that result from the tiebreaking #his preser$es the
stability and the strategyproofness of the student proposing deferred algorithm 'owe$er, tiebreaking
introduces artificial stability constraints Csince, after tiebreaking, schools appear to ha$e strict
preferences between students for whom they are indifferent, and these constraints can harm student
welfare
'47 TE NRMP AL5ORITM
Analysis of the algorithms used by the -ational esidency !atching *rogram clearinghouse showed
that they were closely related to the Gale%hapely algorithm Coth 1==2 #he Gale %hapley result
showed that the -!* $ersion of the algorithm, because it had the hospitals proposing to the
students, produced a match that was hospitaloptimal #he hospital optimal stable matching is
produced by a Mhospital proposingN algorithm, which operates by ha$ing residency programs propose
Cmake offers to applicants and allowing applicants to hold at any point in the algorithm the most
preferred offer amongst those so far recei$ed #he algorithm was criticised as being biased towards
residency programs at the e"pense of the applicants In contrast, defenders of the -!* were
inclined to $iew the matching algorithm as ha$ing e$ol$ed from the traditional recruitment process in
which programs offer positions to applicants, and adapted to changes in the medical market, picking
up special features as required In the fall of 1==7 an alternati$e design of a new Capplicantproposing
-
8/17/2019 Zucas Final
18/69
algorithm was commissioned for conducting the annual match and a study to compare it with the
current -!* algorithm CAl$in+oth, 1==7 According to the same author, the applicant proposing
algorithm was designed based on the algorithm outlined by oth and ande ate and also on the
components of the e"isting algorithm #o code a working algorithm, choices had to be made
concerning the sequencing of operations +ach decision can be shown to ha$e no effect on the
outcome of simple matches but, but could affect the outcome when match $ariations are present Csince
optimal stable matches may not e"ist #he matches produced under the newly designed algorithm
and the current program proposing algorithm were compared and showed that there was a slight
difference between the two algorithms were only 21 of the applicants were affected by the change
in algorithms #he choice between either the applicant proposing or the program proposing was
therefore seen as not in$ol$ing consequential differences for the match as a whole C*eransin and
andlett 1==:
#he goal of the redesign of the -!* was to construct an algorithm that would produce stable
matching as fa$ourable as possible to applicants, while meeting the specific constraints of the medical
market Coth and *eranson 1=== #he comparisons between the new and the e"isting algorithms
were to focus both on how many applicants and residency programs could be e"pected to recei$e
morepreferred and lesspreferred matches under the two algorithms
#he e"istence of married couples amongst the medical students posed some problems as many of the
couples felt that the -!* clearinghouse was not ser$ing them well oth C1=;6 argues that this was
because the matching algorithm used until then did not allow couples to appropriately e"press
preferences 'e went on to clarify that in a market without couples, the 1=72s clearinghouse algorithm
is equi$alent to the deferred acceptance algorithm of Gale and %hapley #he algorithm often fails to
find a stable matching when there are couples in$ol$ed oth further argued that the main problem of
the mechanism is that Cprior to the 1=;5 match it did not allow couples to report preferences o$er
pairs of positions, one for each member of the couple Coth and *eranson 1=== describe the current
algorithm, which elicits and uses couples> preferences o$er pairs of positions and has been used by
more than 62 centralised clearinghouses including the American labour market for new doctors, the -!*
oth C1=;6 postulates that the problem is difficult e$en if couples are allowed to e"press their
preferences o$er pairs of positions, because there does not necessarily e"ist stable matching with
couples In the late 1==2>s, the market e$ol$ed from a decentralised one Coth and Ping, 1==: to
one employing a centralised clearinghouse, where a key design issue was whether it would be
possible to accommodate the presence of couples eilin C1==; reported that under the old
decentralised system, couples had difficulties coordinating their internship choices In 1===, clinical
-
8/17/2019 Zucas Final
19/69
psychologists adopted a centralised clearinghouse using an algorithm based on Al$in oth and
*eranson, in which couples are allowed to e"press preference o$er hospital pairs
+"isting studies on couples matching are mostly negati$e. oth C1=;6 and unpublished work by
%otomayor show that stable matching does not e"ist where there are couples, and onn C1==2 showsthat it may be computationally hard to determine if a stable matching e$en e"ists %ome -!*
participants who participate as couples are ad$ised to form preferences by first forming indi$idual
rank order lists after inter$iewing with programs #hen this list will ser$e as an input into the joint
ranking of the couple For instance, medical students who are couples at the /ni$ersity of ansas
!edical %chool are suggested to make a list of all possible program pair combinations from both
indi$idual rank order lists by computing the difference between the ranking number of the program on
each indi$idual>s rank order list and trying to minimise the difference in their joint rank order list
Cojima, *athak and oth 4212
'48 SUMMARY O9 LITERATURE REVIEW
#he e"istence of the instability in the deferred acceptance algorithm then lea$es a compelling question
as to whether a stable matching can in actual fact ne$er e"ist as implied by the researchersH inability
to pro$e, as done for simple markets, that such a stable matching can e"ist *eculiar and irregular
points are noted which are to not be o$erlooked &hilst the great work these researchers ha$e put in
attempting to pro$e their findings is affirmed, pertinent issues are raised that may in$alidate parts and
not all of their findings *erhaps this can e"plain and enable the construction of a better method to
achie$e the best matching for uni$ersity admissions in (imbabwe From the researches referenced
abo$e, the following points were noted.
i !ost importantly, the deferred acceptance algorithm is processed in iterations which themsel$es
are not e"hausti$e in their e$aluation of criteria and preferences this can be pro$ed this by how
the order of processing matching easily affects the outcome of the matching
ii #he $ery fact that there e"ists proposeroptimal and acceptoroptimal $ersions of the algorithm,
gi$en by which set is made to do the actual MproposalN suggests and indicates failure in the
algorithm to process preferences and criteria in a fair, nonfa$ouring manner for any order of
processing
iii It is noted that, the deferred acceptance algorithm itself is a suggesti$e algorithm, in criteria for
uni$ersities and preferences for students in that it imposes conditions which themsel$es may be
regarded McriteriaN
i$ It should also be noted that in reality, not all problems ha$e the rigid form of preferences, as
imposed by the condition of the original and then later re$isions of the deferred acceptance
algorithm -ot all preferences are made in strict lists some may just not be rigid #he couplesH
problem is an e"ample of nonrigid preferences, howe$er, the past researchers ha$e settled on
calling this a $ariation of the algorithm where couples e"ist
-
8/17/2019 Zucas Final
20/69
'4; CONCLUSION
#he author has managed to analyse the origin of the problem of stable marriage problems, and looked
at the $aried propositions made in attempt to get a solution to this In this case, the researcher aims to
use this to find a suitable solution for the design of a suitably functional centralised uni$ersityadmissions system for uni$ersities in (imbabwe &hilst the fundamental work done in achie$ing
stable marriages in simple markets by Gale and %hapley C1=84 is acknowledged, and the subsequent
re$isions by oth C422:, some anomalies were found that propose and suggest a look at the
algorithm, if possible change and make it suitable for the proposed en$ironment, or if need be,
construct and implement another proposition for the solution of stable matching and use in the design
of a central uni$ersity admissions system
-
8/17/2019 Zucas Final
21/69
CAPTER TREE: METODOLO5Y
)4* INTRODUCTION
#his chapter includes a re$iew of the research method and design appropriateness, a discussion of the
indepth look into the methods used to accomplish the stated objecti$es, including the tools that will
be used to do so
)4& RESEARC DESI5N
esearch design can be defined as a plan that describes how, when and where data is to be collected
and analysed for the purposes of a research C*olit, DF, 3eck, )#, 'ungler, 3* 4226 #his can be
thought of as the logic of a research that throws a light on how the study is to be conducted It shows
how all the major parts of the research study work together in an attempt to address the research
questions #he key to a successful project is proper design and the design stage in$ol$es coming up
with the different modules of the system and their intended functionality
%ystem design can be thought of as the process of defining the architecture, components, modules and
data for the system to satisfy specified requirements #he major objecti$e of this stage is to ensure that
an efficient, effecti$e, maintainable, and reliable system is de$eloped #he system interfaces should be
designed with the end user in mind making the design process easier and the designed work easier to
implement
For the design of the central uni$ersity admissions C)/A model, the researcher dwelled on the
qualitati$e research approach which is an approach to research using methods such as participant
obser$ation, inter$iews, questionnaires or case studies which result in a narrati$e account of a setting
or a practice, with focus being on the interpretation of phenomena in natural settings to make sense in
terms of the meanings people bring to these settings CDenEin, - 4212
#o come up with the model for the centralised uni$ersity admissions system, the researcher had to
re$iew some of the models and their algorithms in literature and how they were designed #he
functionality of these systems was analysed in order to come up with a design that would eliminate
some of the problems that were found to e"ist with these systems A re$iew of the literature on
centralised candidate selection systems therefore aided as a guide in the design of the centralised
uni$ersity admissions system
#he researcher also conducted some inter$iews with the personnel in the 3/%+ admissions offices
with the aim of attaining a deeper understanding of how the admission process is carried out for
undergraduate students, and also the fundamental student entry data required for admission and the
criteria used for selection #his helped in coming up with a model that would incorporate most of the
-
8/17/2019 Zucas Final
22/69
important details and procedures used for the selection of undergraduate students %uggestions and
ideas concerning the $ital features that a centralised admission system should consist of were also
taken from the admissions personnel for the model design
%oftware performance tests were carried out as a way of e$aluating the system>s performance in itsintended en$ironment Different types of tests were conducted, each test gi$ing an e$aluation of the
system under different scenarios For the e$aluation of the system>s attributes that could not be
directly determined by system performance tests, and the effecti$eness of the system, the users>
perceptions had to be taken into account and this was done by collecting the data from two groups of
the system users namely the applicants and the faculty administrators An analysis of data from these
questionnaires was then used for the system e$aluation, thus the e$aluation of the system was a
combination of system tests and users> perceptions
)4& SO9TWARE DESI5N
)4'4& De.i+# Too0.
*roper de$elopment of an ideal system that meets specified requirements and that suits the area in
which the system is to be used in$ol$es the use of appropriate tools In the researcher>s quest to
achie$e the first objecti$e, the researcher decided to use the following tools.
1 *'* scripting language
4 )odeIgniter Framework
5 Internet Information %er$ices CII%
6 3oot%trap for user interface design
7 !y%@< database
PP Scripti#+ L2#+!2+e
#his is a ser$erside scripting language designed for web de$elopment *'* was chosen for the
de$elopment of the system mainly because it offers security on accessing databases and other secure
information, and also that it allows the quick de$elopment of databasedri$en websites
Co"eI+#iter 9r2$e
-
8/17/2019 Zucas Final
23/69
the !odeliew)ontroller C!) software architectural pattern which allows the attaching of
multiple $iews to a model to pro$ide different presentations
I#ter#et I#=or$2tio# Ser3ice.
Internet Information %er$ices ; is a web ser$er software package de$eloped by !icrosoft and this was
adopted as the web ser$er for the centralised uni$ersity admissions system as it tends not to add any
e"tra weight on the machine since it comes with the operating system
BootStr2p
#his is an opensource framework designed to build user interface components 3oot%trap was used in
the designing of the interfaces for the centralised uni$ersity admissions system because it is constant
and ensures that results are uniform across platforms therefore output remains the same whether the
users are using different browsers
MyS>L D2t2%2.e
#his is a relational database management system CD3!% which was adopted for its typical use for
web application de$elopment Coften accessed using *'* thus making it the perfect candidate for
use Also, it generally offers fewer features than other databases, which then means that it is fast and
due to its speed, the system will o$erall be fast
)4'4' Sy.te$ De.i+# Proce..
)4'4'4& Re?!ire$e#t. Speci=ic2tio#
#his refers to the ser$ices that the system will be e"pected to pro$ide for to its users #he
requirements will be classified according to each user
Sy.te$ !.er.:
Applicants
Faculty Administrators
%uper Administrator
)4'4'4' 9!#ctio#20 Re?!ire$e#t.
#hese are statements of ser$ices the system should pro$ide, ie how the system should react to
particular inputs #his is primarily what the system is e"pected and should do.
Q #he system should be able to register new faculty administrators and applicants
-
8/17/2019 Zucas Final
24/69
Q #he system should allow the faculty administrators to add new andRor edit degrees and their
selection criteria
Q #he system should allow all the faculty administrators to $iew the list of selected students
when the selection process is done
Q #he system should allow the super administrator to initiate new intakes
Q #he system should allow the super administrator to add new faculty administrators
Q #he system should allow applicants to place as many preferences as they want upon
application
Q #he system should be able to $erify applicant>s data before sa$ing the application
Q #he system should be able to allocate places to applicants
)4'4'4) No#@9!#ctio#20 Re?!ire$e#t.
#hese specify the criteria that can be used to judge the operation of the system ie how the system is
supposed to be CGlinE, ! 422: #he following are the nonfunctional requirements of the central
uni$ersity admissions system.
Q *erformance requirements
Q Accessibility requirements
Q A$ailability requirements
Q %ecurity requirements
Q %calability requirements
)4'4) So=t
-
8/17/2019 Zucas Final
25/69
Also, this methodology was adopted because it allows the easy implementation of changes to the
system being de$eloped 3ecause there will always be the need to either subtract or add some features
and functionalities to the system in the course of its de$elopment as a way of ensuring the
achie$ement of the stated objecti$es of the research, the agile methodology pro$ed useful due to its
pro$ision of the ability to easily implement new changes
)4'4 Met,o". !.e" i# t,e .y.te$ "e.i+#
Object Oriented Analysis and Design COOAD is a technique for analysing the requirements and
creating the design for a software system #he analysis done in using OOAD needs to be depicted
through $isual modelling and these $isual models of a software system can be created using the
/nified !odelling
-
8/17/2019 Zucas Final
26/69
Figure 1 . )entralised /ni$ersity Admissions %ystem use case diagram
)4'47 Acti3ity Di2+r2$
#his is one of the important /!< diagrams used to describe the dynamic aspects of the system An
acti$ity diagram can be defined as a representation of the flow from one acti$ity to another acti$ity
where an acti$ity can be described as an operation of the system #he author decided to use an acti$ity
diagram for the purpose of capturing the dynamic beha$iour, and the sequence from one acti$ity to
another of the central uni$ersity admissions system #he flow between the system and the actors
within the use cases is described with the acti$ity diagram 3elow is an acti$ity diagram for the
central uni$ersity admissions system
-
8/17/2019 Zucas Final
27/69
-
8/17/2019 Zucas Final
28/69
)4'48 Proce.. 90o< C,2rt
A process flow chart pro$ides a $isual representation of the steps in a process, thereby gi$ing a clear
understanding of the process In this case, the author decided to use a process flow chart as a way of
communicating how the central admissions system would process an application from the time it is
placed on the system to the final stage where a place is allocated to an applicant All the steps that will
be taken by the system and certain decisions to be made during the process will be represented in the
process flow chart #he figure below shows a process diagram for the centralised uni$ersity
admissions system
-
8/17/2019 Zucas Final
29/69
Figure 5 )entralised Admissions %ystem process flow diagram
-
8/17/2019 Zucas Final
30/69
From the diagram abo$e, upon picking an application, the $ery first decision that the system makes is
whether that particular applicant has already been allocated a place in any other uni$ersity #his
decision is the stage where the system aims to eliminate redundant place allocation It is only after the
system determines that the applicant has indeed not been allocated a place, that it considers whether
the applicant>s qualifications meet the stated selection criteria of their first preference chosen If the
applicant does not meet the selection criteria of the preference placed on the first place, the system
goes on to check whether the applicant meets the criteria of the other preferences placed, in their
order In the case that the applicant does not meet the selection criteria for any of hisRher selected
preferences the applicant is not allocated any place
#he other decision that the system makes in the course of the selection process is whether the class
being applied for is full #his decision is only made in the tide that the applicant does meet the
selection criteria of one of the preferences placed Gi$en that the ma"imum number of students that
can be accepted for that particular class has not yet been reached, the applicant is allocated the place
Also in the selection procedure, there is a displacement scenario #his happens when an applicant
meets the selection criteria for a particular preference but the class ha$ing been full In this case, the
system considers if this current applicant is more preferable to any of the applicants making up the
full class %hould the current applicant be more preferable, sRhe is allocated the place of the other
applicant to whom sRhe is preferred to
54; Database Design
As highlighted earlier in this chapter, the !y%@< database, which is a relational database, was
adopted to work with the central uni$ersity admissions system #he following are the reasons which
led to the use of a relational database.
i It offers better security by splitting data into tables thereby making certain tables confidential
ii It caters for future requirements by ha$ing data held in separate tables thereby making it
easier to add records that are not yet needed but may be in the future
iii It is good in handling data through the support for di$erse data needs
3elow is an entity relationship diagram showing the structure of the central uni$ersity admissions
database.
-
8/17/2019 Zucas Final
31/69
Figure 6 )entralised /ni$ersity Admissions %ystem entity relationship diagram
54= /ser Interface Design
#he design process for the user interface calls for a software de$elopment model that suitably suits
websites at the same time making it possible to be integrated with the required components /ser
interface design is concerned with describing user beha$iour and defining how a particular system
will accommodate and respond to that beha$iour C9esse 9ames Garrett, 4211, where user interface is
-
8/17/2019 Zucas Final
32/69
that part of a computer system which a user interacts with in order to undertake certain tasks and
achie$e some particular goals C%tone, 9arrett et al, 4221 It in$ol$es researching into the beha$iours
and goals of the target users of a software system #o ensure that the central uni$ersity admissions
system is well communicated to the users, the author applied the following aspects in the /I design.
i %implicity
#his is one of the core principles that the author applied in designing the /I for the system #he /I of
the central uni$ersity admissions was designed to be as simple as possible so that the intended users
of the system would be able to interpret the design, thus being more inclined to use it #he interface
was designed in such a way that it is clearly laid out, well organised and all the controls that need to
be accessed by the system users are as $isible as possible
ii %tructure
#he author ensured that the interface is clearly laid out, well organised and controls are easily
identifiable for the users #he structure of the /I was constructed with the consideration that the users
would be new to the system and they will try out different options in order to achie$e what they
desire In doing so, they can make a lot of mistakes they can enter wrong data or go to a wrong place
#o minimiEe such, the structure of the /I was considered to play a $ital role
iii %iEe
#his is also an important consideration made in coming up with the /I for the system because the siEe
of the /I directly influences the time taken to load pages, hence the siEe of the /I was minimiEed so
as to make the o$erall system $ery light so that it can respond quickly to the user>s requests
#o be able to come up with a user interface which incorporates the abo$e aspects, the diagram below
shows the sequence of steps taken by the researcher in designing the /I for the central uni$ersity
admissions system.
-
8/17/2019 Zucas Final
33/69
Q Outline%ystemequirements
Q Determinesystemend users
/seresearch
Q /ser *ath!apping
Q %tructureL
-a$igationDesign
Information
Architecture Q Design&ireframe
Q InteractionDesign
Interaction
Design
Q
-
8/17/2019 Zucas Final
34/69
undergoes as it processes an application from the time it is submitted up to when places are allocated
and the merit list generated
554 Applicant>s +nd
#he applicants can interact with the system after logging in, where the applicant would pro$ide user
credentials, or in the e$ent that the applicant does not ha$e an e"isting account, they will ha$e to
create one on the system ha$e to create one on the system
Figure 8 Account creation page for applicants
As depicted abo$e, an applicant>s personal details will be captured into the system during
account creation such that when the applicant logs in, they need not to enter their personal
details, e"cept for their qualifications Once logged in, an applicant will ha$e to create a
profile for hisRher qualifications, before starting the application process #he profile created
would ha$e the qualifications to be used in the selection process
-
8/17/2019 Zucas Final
35/69
Figure : +ntering qualifications for an applicant
-
8/17/2019 Zucas Final
36/69
Once they are logged into the system, they can then select the degree programmes they wish
to apply for in any of the uni$ersities, in order of preference from the list of a$ailable degree
programmes on the system
Figure ; making an application on the system
As depicted abo$e, an applicant makes an application by selecting their degree of choice
through the Mselect degreeN drop down menu and putting the degree at their preferred rank
order Also, an applicant has the ability to change their preference order before the selection
day After submitting their applications on the system, which is only possible before the
application due date, applicants will await notification by the system through an email, of the
a$ailability or una$ailability of an admission place after the selection process is done by the
system
)4)4) 92c!0ty A"$i#i.tr2tor. E#"
#hese will also interact with the system after login #his end of the system will allow each
faculty of e$ery uni$ersity to add degree programmes on offer, thereafter setting the selection
criteria for the degrees in the system before the selection day #he criteria entered in the
system will be the ones which the system would use for the selection of applicants
-
8/17/2019 Zucas Final
37/69
Figure = %etting degree selection criteria
#he faculty administrators will also be in a position to edit or update their selection criteria in
the system whene$er need be After the selection process is carried out by the system, this
end of the system will also show the merit list
)4)4 S!per A"$i#i.tr2tor. E#"
#his end of the system allows the user to set the applications due date initiate a new intake in
the system After the set due date, the system will no longer recei$e applications #he super
administrator will also be responsible for initiating the system to carry out the selection
process after the application due date #he management of uni$ersities in the system will also
be carried out by the super administrator
-
8/17/2019 Zucas Final
38/69
-
8/17/2019 Zucas Final
39/69
Figure 12 !anaging uni$ersities in the system
As depicted abo$e, this end allows the super user to add or remo$e uni$ersities to the system
3esides adding or remo$ing uni$ersities, the super administrator also adds new faculty
administrators to the system
Figure 11 Adding a new faculty
As depicted abo$e, the system administrator adds new faculty administrators to the system as
per uni$ersity
)4)4( Sy.te$ Sec!rity
Any software system de$eloped should be able to enforce non functional requirements such
as integrity preser$ation, accountability, confidentiality amongst other things C&ilhelm et al
4228 #herefore, to ensure that the system meets such requirements, inbuilt security
measures will ha$e to be enforced during the de$elopment of the system In the design of the
central uni$ersity admissions system, the de$eloper ensured that all tables would not be
editable and also that all the user input data is $alidated before being passed on to the system
so as to make sure that the system does not crash as a result of erroneous input Also, access
le$els to different actions which are needed in the system were created, the super and the
-
8/17/2019 Zucas Final
40/69
faculty administrators #his was done to ensure that no unauthorised users will ha$e the
ability to perform certain actions on the system
Another security aspect which was implemented in the de$elopment of the system is the
)ompletely Automated *ublic #uring test to tell )omputers and 'umans Apart Ccaptcha
3ecause the central uni$ersity admissions system is a web based system, the researcher saw it
fit to implement an aspect that would secure it from Distributed Denial of %er$ice CDDo%
attacks, where attackers can make use of some automated software to generate a massi$e
quantity of requests thereby causing a high load on the system ser$er, resulting in legitimate
users and their requests being denied access to the system
)4)47 Se0ectio# A0+orit,$
+ach of the uni$ersities> faculty administrators will set the criteria used to select students for
each of the degree programs offered by the uni$ersity and updating the criteria from time to
time On each of the application submitted on the system, the system checks if the
qualifications of an applicant match or meet the selection criteria for the preferences the
applicant would ha$e put #he steps below gi$e an outline of how the selection algorithm will
process the selection of candidates.
Step 1: #he algorithm takes all the preferences for each student and takes the uni$ersities>
preferences and conditions Cconditional preferences plus criteria
Step : It then chooses the ne"t acceptor Cstarting with uni$ersity 1, / 1, choosing any
acceptor not yet tra$ersed otherwise it will stop processing
Step !: #he current acceptor Cuni$ersity must tra$erse each of its applying students
regardless of the option number where they are applying
a &here the current acceptor is not highest on the current applicant>s rank order list,
acceptor must probe all higher acceptors for tra$ersal of this particular application of
which if all higher precedent acceptors reject current applicant, then that current
acceptor is gi$en an opportunity to accept the applicant Otherwise any of the higher
precedent acceptors of this current application may successfully accept offer and
render current applicant takenRaccepted
-
8/17/2019 Zucas Final
41/69
b Once all applicants ha$e been tra$ersed by current acceptor whether class is full or
not, tra$ersal for the current acceptor is stopped and the algorithm starts again at step
4
)4)48 Po.t Se0ectio#
After the system carries out the selection process, each applicant is notified of the a$ailability
or una$ailability of admission places #he system will also generate a merit list for the
uni$ersities> faculties to see which students will ha$e been admitted to their uni$ersity
%tudents who would ha$e been offered places will go to the uni$ersity they would ha$e been
accepted and hisRher name will be checked for on the merit list
)4)4; Sy.te$ te.ti#+ 2#" e320!2tio#
%oftware testing can be defined as the process of $alidating and $erifying that a software
product meets the technical requirements that guided its design and de$elopment, and also
that it works as e"pected C'asling et al 422; %oftware testing also in$ol$es $erifying to see
if the software product can be implemented with the same characteristic In the researcher>s
quest to achie$e her second objecti$e, the researcher carried out the testing of the centralised
uni$ersity admissions system using benchmarks and also conducted an analysis of the users>
perceptions for the e$aluation of some of the system>s attributes #he findings of the
e$aluation are in )hapter 6 of this document
)4 POPULATION AND SAMPLIN5
#he researcher adopted a sampling strategy known as probability sampling #his is a simple
random method in which students were selected as the applicants to the system #his has an
ad$antage of minimiEing bias by ensuring that each indi$idual has an equal chance of
participating in the research A target population of fifty part 11 students was selected from
the 3indura /ni$ersity of %cience +ducation 3achelor of %cience education degrees in
chemistry and also in physics as the applicants to the system #he researcher selected part 11
students as their appreciation of using the centralised admission system would be based on
prior e"perience with the current decentralised system of uni$ersity admissions and at the
same time, the group will also resemble high school students who would be the actual
intended users of the system since these students were post high school students
-
8/17/2019 Zucas Final
42/69
#he researcher also selected fifteen faculty administrators from the faculties of commerce,
agriculture and en$ironmental science and the faculty of commerce for each of the
uni$ersities within the scope of the system which are 3indura /ni$ersity of %cience
+ducation, /ni$ersity of (imbabwe and the -ational /ni$ersity of %cience and #echnology
A mock application e"ercise was performed for the selected sample of fifty applicants where
the applicants applied to each of the three uni$ersities for degrees of their choice #he faculty
administrators were also gi$en an opportunity to perform the administrati$e tasks on the
system After using the system, the students and the faculty administrators ga$e their $iews
and perceptions on the centralised uni$ersity admissions system
)4( DATA ANALYSIS PROCEDURES
#his refers to the steps followed in data analysis From the number of different data analysis
software a$ailable that can be used to come up with a conclusion on the $alue of a research,
the researcher will adopt the %tatistical *ackage for the %ocial %ciences C%*%% #he
researcher chose %*%% as the data analysis tool due to the wide range of options that it offers
ie it is specifically made for analyEing statistical data and thus it offers a great range of
methods, graphs and charts Also, %*%% pro$ides better output organiEation since it is
designed to make certain that the output is kept separate from the data itself All results are
stored in a file that is different from the data
-
8/17/2019 Zucas Final
43/69
CAPTER 9OUR: DATA PRESENTATION AND ANALYSIS
4* I#tro"!ctio#
#his chapter takes a look at the e$aluation of the centraliEed uni$ersity admissions system through
$arious system tests and the analysis of user perceptions #he usability of the central uni$ersity
admissions model in its intended en$ironment was e$aluated based on the following software
attributes which are scalability, stability, and speed Different tests to e$aluate the performance of the
system were carried out #he system users> perceptions were analysed to gi$e an e$aluation of the
system based on those attributes that could not be unco$ered through system tests and also its
effecti$eness
4& A#20y.i. 2#" I#terpret2tio# o= Re.!0t.
4&4' Sy.te$ Per=or$2#ce Te.t.
#he system performance tests were carried out for the purpose of testing whether the system would be
able to cope with the different $ariant en$ironmental factors that may make it redundant nown tests
in software engineering under the field of system performance testing were used for the purpose of
determining how the system will perform in terms of responsi$eness and stability under a particular
workload
#he researcher used performance testing for the measuring of other quality attributes of the systemsuch as scalability, reliability and resource usage 3eing a subset of performance engineering,
performance testing is also regarded as a computer science practice which stri$es to build
performance into the implementation, design and architecture of a system
For this research, the most common performance concerns related to a web application were adopted
and these are the speed of the central uni$ersity admissions system, the ability to accommodate the
current or e"pected user base, the ability to handle stress and the ability to perform within capacity
For the testing of the system, the researcher used two different laptop machines as the ser$ers #he
machines had the following specifications.
T2%0e & 2r"
-
8/17/2019 Zucas Final
44/69
4&4'4& St2ti.tic. =or t,e t2r+et e#3iro#$e#t
#he proposed system will be run on the web, and accessible from anywhere regardless of
geographical location According to the -o$ember 4216 A>
-
8/17/2019 Zucas Final
45/69
which means
the speed of
the system is
within the
normal '##*
bounds ie
822
milliseconds,
!icrosoft
Appendi"
*'* or
!y%@<
software
*'* or
!y%@<
software
!y%@<
software
#his was
the
probable
cause of the
two '##*
725 errors
%calability
Interpretatio
n
Ser3e
r &
All
successful
'##* 422
responses
All
successful
'##* 422
responses
All
successful
'##* 422
responses
Ser3e
r '
All
successful
'##* 422
responses
All
successful
'##* 422
responses
All
successful
'##* 422
responses
equests for
both ser$ersturned
around
without a
timeout or
ser$ice
una$ailable
esource
demand alsodid not
increase any
significantly
for both
ser$ers for
the II%, *'*
or !?%@<
software
esource
demand alsodid not
increase any
significantly
for the II%,
*'* or
!y%@< for
both ser$ers
From the table abo$e, the results of the speed test performed show that the system managed to sustain
o$er the time gi$en, with only minimal lag e"perienced when the number of requests increased,
considering the fact that the actual recommended hardware is usually more powerful than the one
used shown in #able 61
For the hardware scalability test, two different machines with specifications shown in table 61 were
used to assess whether a change in hardware can affect the speed at which the system will perform
-
8/17/2019 Zucas Final
46/69
and as shown in the table, the application maintained its performance e$en when it was run on a
machine of lower hardware capability All the requests for %er$er 4 pro$ed successful and no
significant resource usage #he speed and the scalability of the system can therefore be considered to
be good from the results contained from the performance tests
4&4'4) A""itio#20 Te.t.
#hese are tests which were conducted to determine other characteristics of the system which could not
be determined through the system performance tests #hese factors had to be tested for in the system
to further determine its usability in its en$ironment
T2%0e ) A""itio#20 .y.te$ te.t.
Te.t Re.!0t I#terpret2tio#
Te$per 9ree Good 3ecause the system is centrally
managed, indi$idual uni$ersities
and faculties ha$e no
manipulati$e control o$er who
they accept e"cept those that
qualify by merit of that
uni$ersityHs criteria
Ce#tr20i2tio# ery Good &eb Accessible and centrally
processing applications
No#@re"!#"2#t ery Good 3ecause no one candidate can
recei$e a further offer after
ha$ing recei$ed one, it means
no candidate can recei$e two
offers at any gi$en time
!$2#@Re.o!rce
Mi#i$20
ery Good +"cept for the inputting of
criteria L preferences and also
setup data the system will runwith $ery minimal human
inter$ention
#he test for the humantemperfree feature carried out on the system show that the selection outcome
from the centralised uni$ersity admissions system cannot be manipulated or influenced by users since
the system is centrally managed which lea$es uni$ersities and faculties with no control o$er who is
accepted #he system passed the test with a good score #he system also passed the test for the
centraliEation feature with a $ery good score due to its web accessibility #he other additional test
-
8/17/2019 Zucas Final
47/69
which was carried out on the system was the test for nonredundancy which the system passed with a
$ery good score as the test resulted in no applicant being allocated more than one uni$ersity place A
test was also carried out to assess whether the centraliEed uni$ersity admissions system is human
resource minimal which the system passed with a good score because besides the setting of the
selection criteria, the system can carry out the selection process on its own with $ery little human
assistance
4&4) A#20y.i. o= t,e ?!e.tio##2ire.
#he analysis and interpretation of the results yielded two sets of data that were also used to reach a
conclusion of this research thesis, in addition to the system tests carried out #hese two sets of data are
non numerical and numerical data -onnumerical data is the data that came as a result of the
questionnaires pro$ided to the fifty applicants and fifteen faculty administration officers who took
part in the research #his nonnumerical data was used to create the numerical data which was
represented in the form of frequency tables as will be shown in this chapter #he frequency tables
show the occurrences of certain responses, and these frequencies are the ones which were also used
for representing data in the form of percentage bar charts #he researcher collected data from 7;
participants with 68 being potential applicants and the remaining 14 being the faculty admission
administrators #wo sets of questionnaires were used in the research with one set being for the
applicants and the other being for the faculty administrators #he e$aluation of the usability of the
centralised application system was achie$ed through an analysis of the data obtained from these
questionnaires #he samples of the questionnaires are included in the inde"
4&4)4& A#20y.i. o= t,e App0ic2#t >!e.tio##2ire
#he Applicant @uestionnaire was used in this research for the purpose of gathering the users>,
particularly the applicants, perceptions on the usability of the centralised uni$ersity admissions
system thereby imparting a basis of e$aluating the system>s usability #his questionnaire includes
questions that focus on determining the applicants> $iews and opinions regarding the usability of the
centralised uni$ersity admissions system and also related factors to the system
-
8/17/2019 Zucas Final
48/69
>N I ,23e prior eperie#ce
-
8/17/2019 Zucas Final
49/69
Frequency *ercent alid *ercent
)umulati$e
*ercent
alid ?es 62 ;:2 ;:2 ;:2
-o 8 152 152 1222
#otal 68 1222 1222
From the frequencies shown in the table abo$e, it can be concluded that most applicants will be able
to access the system since it is web based thus widening its accessibility
>N I =o!#" t,e ce#tr20i.e" !#i3er.ity 2"$i..io#. .y.te$ re2.o#2%0y e2.y to !.e
T2%0e ( I =o!#" t,e ce#tr20ie" .y.te$ re2.o#2%0y e2.y to !.e
Frequency *ercent alid *ercent
)umulati$e
*ercent
alid %trongly Agree 16 526 526 526
Agree 48 787 787 ;:2
Disagree 6 ;: ;: =7:
%trongly
Disagree4 65 65 1222
#otal 68 1222 1222
-
8/17/2019 Zucas Final
50/69
9i+!re &) t,e ce#tr20i.e" 2"$i..io# .y.te$ i. re2.o#2%0y e2.y to !.e
In order to determine whether the design of the centraliEed admissions system enhanced ease of its
use, the researcher had to assess if the applicants found the system easy to use from the e"perience
they had with it As shown in the tabulated statistics abo$e, 52 of the applicants strongly agreed that
they found the )/A system relati$ely easy to use, with 7: also agreeing on the ease of use of the
system 'owe$er, 15 of the respondents did not agree to this, with 6 of this strongly disagreeing
esponses on this question can conclude that the centralised uni$ersity admissions system is quite
usable
-
8/17/2019 Zucas Final
51/69
>N( I =o!#" c2rryi#+ o!t t,e 2pp0ic2tio# proce.. e2.y
9i+!re & C2rryi#+ o!t t,e 2pp0ic2tio# proce..
-
8/17/2019 Zucas Final
52/69
of the system was kept as simple as possible resulting in the conclusion which can be drawn from the
gi$en responses that the system is simple
9i+!re &( t,e .y.te$ i. !##ece..2ri0y co$p0e
>N8 T,e ce#tr20i.e" 2"$i..io#. .y.te$ i. co.t e==ecti3e
-
8/17/2019 Zucas Final
53/69
Frequency *ercent alid *ercent
)umulati$e
*ercent
alid -o 6 =15 =15 =15
?es 64 ;: ;: 1222
#otal 68 1222 1222
>N; T,e ce#tr20i.e" 2"$i..io#. .y.te$ s ability to eliminate human errors
and bias in the admissions process A total percentage of 48 and 6; were in agreement that the
centralised system will do eliminate human errors and bias, hence fair A total percentage of 44 and 6
were howe$er in disagreement to the $iew because they highlighted that the manual method is really
not biased %ince a significantly greater number of the population agreed that human error and bias
can be eliminated, it can therefore be concluded that the centralised system eliminates human error
and bias in the admission process
4&4)4' A#20y.i. o= t,e 92c!0ty A"$i#i.tr2tor. >!e.tio##2ire.
#his questionnaire was designed for the assessment of the centraliEed admissions system by the
faculty administrators after using it #he questionnaire contains questions aimed at determining the
administrator>s perceptions on the usability of the system and also as a way of attesting other
-
8/17/2019 Zucas Final
54/69
properties of the system such as the redundancy elimination, reduction in the time taken to carry out
the selection process, reduction in the need for human mediation in carrying out the actual selection
process
>N& ,o< +oo" i. t,e i#ter#et co##ecti3ity 2t yo!r !#i3er.ity
#his was asked to determine whether some faculty administrators would not be able to access the
system due to limited or no internet connecti$ity at their uni$ersities Out of the total responses, 5 and
: people said that their internet connecti$ity was e"cellent and good respecti$ely whereas 4 said that it
was poor -one said that they did not ha$e connection at all #his implies that the design of the
centraliEed admissions system as a web based system enables easy access to its intended users as
shown by the greater portion of the people with internet connecti$ity 3elow is a presentation of the
distribution of the responses gi$en to the question.
9i+!re &7 I#ter#et co##ecti3ity
>N' "o yo! ,23e t,e 2%i0ity to 2cce.. t,e p02t=or$ 2. 2#"
-
8/17/2019 Zucas Final
55/69
connecti$ity at their uni$ersity was poor in the pre$ious question A conclusion can therefore be made
that the centraliEed application system is highly accessible since it is web based #he responses to this
question are shown in the pie chart below.
9i+!re &8 A%i0ity to 2cce.. p02t=or$
>N) I =o!#" t,e ce#tr20ie" !#i3er.ity 2"$i..io#. .y.te$ re2.o#2%0y e2.y to !.e
#he faculty administrators were also asked whether they found the system reasonably easy to use in
which 47 of the respondents strongly agreed and 64 just agreeing to that %ome of those who
agreed highlighted that the design of the model was simple, consequently its ease of use Another
47 of the respondents indicated that they did not find the system quite easy to use and the remaining
; strongly disagreeing that they found the centralised admissions system reasonably easy to use
3elow are the presentations of the faculty administrators> responses to this question
T2%0e ; t,e ce#tr20i.e" 2"$i..io# .y.te$ i. re2.o#2%0y e2.y to !.e
Frequency *ercent alid *ercent
)umulati$e
*ercent
alid %trongly Agree 5 472 472 472
Agree 7 61: 61: 88:
Disagree 5 472 472 =1:
%trongly Disagree 1 ;5 ;5 1222
#otal 14 1222 1222
-
8/17/2019 Zucas Final
56/69
9i+!re &; E2.e o= !.e o= t,e ce#tr20i.e" .y.te$
>N7 I =o!#" per=or$i#+ t,e 2"$i#i.tr2ti3e t2.1. .i$p0e
#he table below shows the responses gi$en to this question = of the respondents either strongly
agreed or agreed to ha$ing found the administrati$e tasks simple whereas the remaining 5 strongly
disagreed and disagreed to ha$ing found performing the tasks simple #hose who claimed not to ha$e
found performing tasks simple highlighted that they had some challenges in the setting of the degrees
criteria in the system although they would e$entually manage to do so
T2%0e per=or$i#+ 2"$i#i.tr2ti3e t2.1.
-
8/17/2019 Zucas Final
57/69
Frequency *ercent alid *ercent
)umulati$e
*ercent
alid %trongly Agree 4 18: 18: 18:
Agree : 7;5 7;5 :72
Disagree 4 18: 18: =1:
%trongly Disagree 1 ;5 ;5 1222
#otal 14 1222 1222
>N 8 5i3e 2# o3er200 r2ti#+ o= t,e .y.te$ i# ter$. o= it. !.2%i0ity
As in the applicant questionnaire, the faculty administrators were asked to gi$e their perception on the
centraliEed admission system basing on how usable it is From the results obtained, =4 of the
respondents rated the system as e"cellent, $ery good and just good #he remaining ; responded that
the system was poor, mainly due to the challenges they had faced in setting the degree criteria for
their faculties #his means that o$erall the users appreciated the usability of the system #he
following graph gi$es a summary of the responses to the question.
9i+!re & O3er200 .y.te$ r2ti#+
-
8/17/2019 Zucas Final
58/69
>N; t,e ce#tr20i.e" 2"$i..io# .y.te$ i. 0i1e0y to re.!0t i# t,e re"!ctio# i# t,e 2"$i..io#
re.po#.e ti$e
#he faculty administrator>s perceptions were sought on the speed of the centraliEed admission system
by asking them whether they see the likeliness of a reduction in the admission response time throughthe system where the majority consisting of ;5 of the respondents gladly agreeing that the
admission response time will be reduced, highlighting that manually going through applications
ob$iously takes much time and the remaining 1: disagreeing that the centralised admissions system
will reduce the response time taken in the candidate selection process It can therefore be concluded
that the centraliEed admissions system reduces the time taken in the candidate selection process thus
the admission response time A summary of the results is shown in the following table
T2%0e &* t,e ce#tr20i.e" 2"$i..io# .y.te$ N t,e ce#tr20ie" !#i3er.ity 2"$i..io#. .y.te$
-
8/17/2019 Zucas Final
59/69
9i+!re '* Re"!ctio# i# re"!#"2#cie.
>N&* t,e ce#tr20i.e" !#i3er.ity 2"$i..io#. .y.te$
-
8/17/2019 Zucas Final
60/69
9i+!re '& ,!$2# i#te#.ity re"!ctio#
4' C,2pter S!$$2ry
#his chapter ga$e an analysis of the e$aluation of the system which was done in two parts, the first
part being system performance tests and the other part being that of users> perceptions on the
effecti$eness of the system #he results drawn from the performance tests conducted on thecentraliEed uni$ersity admissions system earlier in this chapter showed the functionality of the system
in different scenarios #his was used to determine the usability of the system in its anticipated
en$ironment
#he sample of results shown in the last part of this chapter analyses the users> perceptions on the
effecti$eness and also properties of the system that could not be drawn from the system performance
tests conducted From these research findings, a greater part of the population from the sampled data
attested to the technical usability and the effecti$eness of the system Also, a greater number of the
administrators who had the opportunity to use the system showed satisfaction on the system>s
effecti$eness in eliminating redundant offers in the candidate selection process and also how it would
speed up the candidate selection process
-
8/17/2019 Zucas Final
61/69
CAPTER 9IVE: CONCLUSIONS AND RECOMMENDATIONS
(4* I#tro"!ctio#
#he researcher had two objecti$es that she intended to meet in the course of this research which were
to design a centralised uni$ersity undergraduate admission system model using the stable matching
problem concept and to test the system model to e$aluate the usability and the effecti$eness of the
resulting design of the centraliEed undergraduate admission system 3asically, this chapter presents a
summary of the findings and the conclusions drawn from the research Future works that can be
de$eloped from this project are also recommended in this chapter
(4& Ai$. 2#" o%jecti3e. re20i.2tio#
#he aim of the project was to de$elop a centraliEed uni$ersity undergraduate candidate selection
system that is capable of allocating uni$ersity places to applicants in a fast, accurate and non
redundant manner #he researcher managed to come up with a centralised undergraduate uni$ersity
admissions system which implements the stable matching concept #he system was e$aluated using
system performance tests and from the results presented in the pre$ious chapter, a conclusion can be
made that the system is technically usable #he effecti$eness of the system was also determined
through user perceptions and the results of the analysis of the user perceptions were also presented in
the pre$ious chapter From those results, it can be concluded the centralised admissions system is
effecti$e in eliminating redundancies in the undergraduate admissions process
(4' 9!t!re Wor1
As the future of this research, the centralised uni$ersity admissions system should be impro$ed so that
besides undergraduate admissions, other admission types such as postgraduate admissions can also be
accommodated in the centralised system Also, applicants with special qualifications should also be
catered for, besides the A>
-
8/17/2019 Zucas Final
62/69
&orks )ited
1 A + oth 1=;4, T#he +conomics of !atching. %tability and Incenti$esT, !athematics of
Operations e$iew, pp 81:84;
4 Abdulkadiroglu A 4227, T)ollege Admissions &ith Affirmati$e ActionT, International 9ournal
of Game #heory, $ol 55, no 6
5 Abdulkaroglu et al 4227, T#he 3oston *ublic %chool !atchT, American +conomic e$iew
6 Atila Abdulkadiroglu L #ayfun %onmeE 4225, T%chool )hoice. A !echanism Design
ApproachT, American +conomic e$iew, $ol =5, pp :4=:6:
7 3iro et al 4212, T#he )ollege Admissions *roblem &ith
-
8/17/2019 Zucas Final
63/69
17 *olit, DF, 3eck, )#, 'ungler, 3* 4226, +ssentials of -ursing esearch. !ethods,
Appraisal, and /tiliEation, 93
-
8/17/2019 Zucas Final
64/69
APPLICANT >UESTIONNAIRE
?our participation is kindly requested in our questionnaire on the uni$ersity candidate selection
system In this questionnaire you are kindly asked to respond to a few short questions regarding your
$iews on the system #hank you so much for your $aluable time and most of all your $ital support andhonest perception
SECTION A: PERSONAL IN9ORMATION
Tic# $here appropriate
1 Gender. male female
4 Age range. 1;47 4852 5162 61and abo$e
SECTION B: INTERNET EPERIENCE
5 I ha$e prior e"perience with the use of the internet. ?es -o
6 Do you ha$e access to a de$ice that can connect you to the internetJ
?es -o
7 Do you ha$e the ability to access the platform from your areaJ
?es -o
SECTION C: ANALYSIS O9 TE USABILITY O9 TE CENTRALIZED APPLICATION
SYSTEM
8 #he centralised admission system is easy to use
%trongly agree Agree Disagree %trongly disagree
: )arrying out the application process was straightforward
-
8/17/2019 Zucas Final
65/69
%trongly agree Agree Disagree %trongly disagree
; I found the system unnecessarily comple"
%trongly agree Agree Disagree %trongly disagree
= I had to be assisted through the application process on the system
?es -o
12 #here is a part of the process I could not complete due to the system>s infeasible requirements
?es -o
11 Gi$e an o$erall rating of the system in terms of its usability
+"cellent ery good Good *oor
SECTION D
14 #he current admission process is subject to bias in the selection process
%trongly agree Agree Disagree %trongly disagree
15 Automated admissions process is likely to eliminate bias in the selection process