Application of Genetic Algorithm in Modeling University Admission Decision Support System

Application of Genetic Algorithm in Modeling University Admission Decision Support System

ABSTRACT

This thesis evaluates the impact of Universities’ entry qualification requirements and the performance of students, in order to decide how these factors bring patterns that are best for admission. The proposed model uses the defined University’s entry qualification as input variables and the students’ first year performance to determine the best admission requirements. Genetic algorithm was used as the searching technique to determine the hidden relationship between the input and the associated performance. Students’ admission data and their corresponding first year results were obtained from the department of Mathematics, Ahmadu Bello University, Zaria. The results indicated that the observed performance of students whose admission into Mathematics Department through the University Matriculation Examinations, Post University Tertiary Matriculation Examinations and O’levels depends more on their respective mathematics and physics average performance in all the three examinations than their entry scores in the individual examination. A comparative study using a statistical model show that the result obtained from the genetic algorithm approach were in line with the result of the statistical model. The model was implemented using java programming language, developed in Net-bean environment.

TABLE OF CONTENTS

DECLARATION …………………………………………………………………………………………… iv
CERTIFICATION …………………………………………………………………………………………… v
DEDICATION ………………………………………………………………………………………………. vi
ACKNOWLEDGEMENT ……………………………………………………………………………….vii
ABSTRACT ………………………………………………………………………………………………….. ix
TABLE OF CONTENT……………………………………………………………………………………. x
CHAPTER ONE …………………………………………………………………………………………….. 1
GENERAL INTRODUCTION ………………………………………………………………………….. 1
1.1 Introduction ………………………………………………………………………………………………. 1
1.2 Background Information ……………………………………………………………………………… 3
1.3. Problem Definition and Motivation ………………………………………………………………. 5
1.4. Objective of the Study ……………………………………………………………………………….. 6
1.5 Research Methodology ……………………………………………………………………………….. 6
1.6 Contribution to Knowledge ………………………………………………………………………….. 7
1.8 Significant of the Study ………………………………………………………………………………. 9
1.9 Organization of the Thesis …………………………………………………………………………… 9
CHAPTER TWO ………………………………………………………………………………………….. 10
REVIEW OF LITERATURE ………………………………………………………………………….. 10
2.1 Introduction …………………………………………………………………………………………….. 10
2.2 History of Evolutionary Algorithms …………………………………………………………….. 10
2.2.1. Search Techniques ………………………………………………………………………………… 12
2.2.2 Evolution Theory and Genetic Algorithm ………………………………………………….. 15
2.3 Genetic Algorithms ………………………………………………………………………………….. 16
2.3.1 Applicability of Genetic Algorithm …………………………………………………………… 20
2.4 University Education in Nigeria ………………………………………………………………….. 21
2.5 Related Work ………………………………………………………………………………………….. 23
xi
2.6 Literature Gap …………………………………………………………………………………………. 31
CHAPTER THREE ……………………………………………………………………………………….. 33
MODELING THE STUDENT’S PERFORMANCE DECISION SUPPORT SYSTEM
…………………………………………………………………………………………………………………… 33
3.1. System Description ………………………………………………………………………………….. 33
3.2. Benchmarking Student’s Performance ………………………………………………………… 34
3.3 Features Extraction and Normalization of Data ……………………………………………… 35
3.3.1 Handling Discrepancy of Feature Extraction ………………………………………………. 35
3.3.2 Handling of Normalization………………………………………………………………………. 39
3.4 Genetic Algorithm ……………………………………………………………………………………. 42
3.4.1. Fitness Measurement …………………………………………………………………………….. 45
3.4.2 Selection ………………………………………………………………………………………………. 46
3.4.3 Crossover …………………………………………………………………………………………….. 46
3.4.4 Mutation ………………………………………………………………………………………………. 46
3.5 Basic Genetic Algorithm Procedure …………………………………………………………….. 47
3.5.1 Initial Population Generation …………………………………………………………………… 47
3.5.2 Fitness Evaluation………………………………………………………………………………….. 49
3.5.3. New Population ……………………………………………………………………………………. 49
3.5.4 Acceptance …………………………………………………………………………………………… 53
3.5.5 Termination Condition ……………………………………………………………………………. 54
CHAPTER FOUR …………………………………………………………………………………………. 55
IMPLEMENTATION OF THE DECISION SUPPORT SYSTEM ………………………… 55
4.1 Introduction …………………………………………………………………………………………….. 55
4.2 System Requirement …………………………………………………………………………………. 55
4.2.1 Hardware Requirement …………………………………………………………………………… 55
4.3.1 The Data Use for Evaluation ……………………………………………………………………. 55
4.3.2 Data Normalization ………………………………………………………………………………… 56
xii
4.3.3 Data Storage ……………………………………………………………………………………….. 56
4.4 System Implementation …………………………………………………………………………….. 56
4.4.1 Upload Unit ………………………………………………………………………………………….. 57
4.4.2 Subjects Selection Unit …………………………………………………………………………… 59
4.4.3 Selection Unit ……………………………………………………………………………………….. 60
4.4.4 Searching Unit ………………………………………………………………………………………. 61
4.5 Result and Discussion ……………………………………………………………………………….. 63
4.6. Validation ………………………………………………………………………………………………. 66
4.7 Conclusion ……………………………………………………………………………………………… 68
CHAPTER FIVE …………………………………………………………………………………………… 69
SUMMARY, CONCLUSION AND RECOMMENDATION ……………………………….. 69
5.1 Summary ………………………………………………………………………………………………… 69
5.2 Conclusion ……………………………………………………………………………………………… 69
5.3 Recommendations ……………………………………………………………………………………. 70
REFERENCES……………………………………………………………………………………………… 71
APPENDIX ………………………………………………………………………………………………….. 80
xiii

CHAPTER ONE

INTRODUCTION

1.1 Introduction

University education in Nigeria has witnessed tremendous development since the country’s independence in 1960 (Adeyemi, 2010). This is in recognition of the fact that the national policy of education stipulates that university education in Nigeria shall make optimum contribution to national development by intensifying and diversifying its programmes for the development of high level manpower within the context of the needs of the nation (FGN, 2004). One of the factors limiting the University in performing its roles as it is required is the quality of students admitted into various academic programmes (Adeyemo and Kuye, 2006). It is expected that an average student admitted into the University should be able to face academic studies with ease and pass his/her courses without engaging examination malpractice; because it is assumed that such student would have had prior experience in public examination.

Students are expected to have sat for the Senior Secondary School Certificate Examinations (SSCE) and passed the minimum requirement and presented themselves for the Joint Admission and Matriculation Board (JAMB) Examination as a selection test and pass at acceptable cut-off point before being offered admission into the university (Salim, 2006). Despite these public examinations that the Nigerian undergraduates had gone through, it has been observed that their performances in the first two years of their undergraduate studies do not usually match that of the JAMB which is used as the basis for their admission in the first place into the University (Adeyemo and Kuye, 2006). Many students hardly pass all their first year courses and majority of those who successfully do so usually have poor grades. A great percentage of university graduates in Nigeria fall below second class upper division and the number of spillover students in various departments are equally high. The situation gets worse as those who manage to graduate are not productive in the labour market because they are unable to meet the expectations of the employers (Ajala, 2010).

In 2005, Universities recommended that further screening be conducted on candidates who sat scored between 180 and 200 marks in the university matriculations examination depending on the university where admission is being sought. The recommendation was made with the hope that the post- JAMB screening exercise would restore the past glory of tertiary education in the country and make university education accessible only to those who want and need it (Ande, 2006). Hence, these lead to multiple admissions selection criteria.

The issue of whether or not the scores in O’levels, UME and PUTME correlate to the candidate’s performance in the university, especially in the first year has begun to attract researchers’ attention. Some researchers claimed that O’levels and UME have no correlation. This needs further investigation and evaluation in order to arrive at a reasonable conclusion. It is known that the selection of students is a complex decision making process, in which multiple selection criteria often needs to be considered.

However, the selection criteria used in higher education admission processes varies widely among programmes and no consistent conclusions can be reached on the predictive values of these criteria (Wilson, 1999). Statistical procedures, such as discriminant analysis and regression analysis are traditionally used for predicting the potential academic success of the applicant (Graham, 1991). In the world of information processing, there are lots of data with increasingly complex multi-domain problems containing either real-world or computer-generated data, which the statistical data processing tools may not be sufficient enough to handle, hence a more advanced approach needs to be developed.

This thesis uses genetic algorithm to evaluate the admission requirement into various university programmes using computer science of the department of mathematics, Ahmadu Bello University Zaria as a case study. The system predicts the patterns that are suitable for selection of students’ into the programme. Genetic algorithm has shown a promising feature in the area of decision support system. The principle of survival of the fittest in which genetic algorithm was modeled, could be of great benefit in the process of random selection from the available data.

1.2 Background Information

Despite stringent measures and strategies employed by the Nigerian government to ensure that educational standards are maintained at least at university level, students who after passing through these vigorous examinations still perform far below expectations. For instance, from the summary of Computer Science students’ data, Mathematics department, Ahmadu Bello University, Zaira for 2009/2010 session, out of 173 students that were admitted into the programme none had CGPA above 4.5, 18 had CGPA between 3.5 and 4.49, 35 had CGPA between 2.4 and 3.49, 10 had CGPA between 1.5 and 2.39. At the end 97 student were recorded to have one or two carry over and 2 were asked to withdraw (Departmental second semester summary, 2010). This implies that only 10.78% of the students actually had satisfactory results at the end of their stay of first academic year. This also shows that 87.22% of the students had academic challenges as undergraduate students. The high rate of poor academic achievement among undergraduate is not unconnected with the channel through which they gained entry into the University. Ebiri (2010), observed that using JAMB as a yardstick for admission of students into Nigerian universities has led to the intake of poor caliber of candidates, characterized by high failure rate, increase in examination malpractice, high spillovers and the production of poor quality output that are neither self-reliant nor able to contribute effectively in the employment world.

Ironically, the process of selecting candidates for admission into tertiary institutions has largely depended on some fixed combinations of some subjects taken by applicants in their lower level classes. However, this technique has never been proved efficient in admitting candidates that may perform well in the chosen courses. The fast growing of candidates seeking for admission into tertiary institutions, there is a need to use past data for decision support in admitting suitable candidate for a course of study.

Universities are facing the immense and quick growth of the volume of educational data (Schönbrunn and Hilbert, 2006). Intuitively, this large amount of raw stored data contains valuable hidden knowledge, which could be used to improve the decision making process of universities (keshavamurthy et al., 2010). An analysis of the existing transaction data provides the information on students that will allow the definition of the key processes that have to be adapted in order to enhance the efficiency of studying (Mario et al., 2010). It is tedious and difficult to analyze such large voluminous data and establishing relationship between multiple features manually. Our proposed system delves into the problem of finding data patterns in admission datasets and provides a technique to predict the performance of students in the first year in the University based on the admission combination.

1.3. Problem Definition and Motivation

Higher education systems all over the world nowadays are challenged by the new information and communication technologies (Boufardea and Garofalakis, 2012).

Moreover, with the increase in competition among the prospective students into higher institutions, most Universities are facing the daunting task of selecting the best students, who have the ability and skills to pursue and succeed in their academic career in a particular field of studies. This is because Universities are interested in increasing performance. Performance is one of the means of measuring University’s quality and reputation (Jusoff et al., 2008), thus higher institutions are becoming more interested in predicting the paths of students, and identifying which students will require assistance in order to graduate (Luan, 2004). In order to be able to achieve this objective, the finding relationships and patterns that exist but are hidden among the vast amount of educational data is needed. This knowledge will help in educational main processes such as counseling, planning, registration, and evaluation in order to give suitable recommendation of the students.

Predictions of qualities of entry result that should be used in admitting students into respective programmes are published in Nigeria, mostly in medicine, education and engineering and most of these are done using statistical approaches. The work of Adewale et al (2007) and Luna (2004) show a great insight that the field of computer science has a lot to offer in contributing to the knowledge evaluation and the effectiveness of JAMB-UME Scores, post-UME scores and SSCE Scores.

The aim of this thesis is to determine how aggregation of UME, post-UME and SSCE scores bring a pattern that is commonly attributed to the good performance of first year students’ academic achievement at the university in the department of Mathematics, Ahmadu Bello University, using the concept of Genetic Algorithm. The identification of these patterns can help in the selection process for admitting students into the various departments.

1.4. Objective of the Study

The main objective of this thesis is to design a model using genetic algorithm that can be employed in searching trends or pattern in student’s previous admission records. This is achieved by using the aggregation of UME, Post UME, O’level scores against their corresponding CGPA at the end of their first academic year in the University. Realizing this objective can help in candidates’ selection criteria for admission process into the university. This main goal can be achieved by means of the following objectives which are:

1. To determine the means by which data collected can be translated to meaningful ones.

2. To develop a model for searching hidden pattern among the available data set using genetic algorithm.

3. To implement model of genetic algorithm

4. To test and validate the model using real data of students’ records.

1.5 Research Methodology

In designing the system, the objectives stated in section 1.4 can be achieved by considering the following steps:

1. Data collected are subjected to the process of feature extraction and normalization. The data gathering process involves the collection of raw data about students, which include the UTME score, PUTME score and O’level results (which are the entry requirements into the University). Feature extraction is carried out since the data collected can be inconsistence, incomplete or noisy.