PREDICTING STUDENT
PERFORMANCE USING NEURAL NETWORK
CHAPTER ONE
INTRODUCTION
1.0 Background of the Study
Education
is an essential factor for improving the socio-economic, cultural, and
political development of a country; for this reason, its role cannot be
overemphasized (Ajayi & Ekundayo, 2008). In
our contemporary world, higher education is an important means for economic and
social development and progress of a country. It should be noted that higher
education is not just one of many means to a middle-class life; it has become
most essentially the only means (Tierney, 2006).
The
university admission system is charged with the responsibility of admitting
prospective students into the university using guidelines that are set by the
Joint Admission and Matriculation Board and the National University Commission.
The Federal government of Nigeria established the Joint Admission and
Matriculation Board (jamb) in 1978 to handle admission processes (Asein &
Lawal, 2007). The board aims at establishing a unified standard for carrying
out matriculation examination and giving admission to qualified candidates into
the university academic system (Asein & Lawal, 2007).
A
prospective candidate must acquire at least a five credit pass in relevant
senior secondary school subjects include Mathematics and English and also to
score the required score in the Unified Tertiary Matriculation Examination
(UTME) for the desired choice of institution and course (Salim, 2006). The
guidelines set by the federal government for admission into institution of
higher learning (State, Federal and Private Universities) are based on quota
systems in which 45% of candidates are admitted on merit, 35% on catchment and 20% on educationally less developed
states mostly the northern states.(Bakwaph, 2013). Considering the statistics
above only 45% of students are admitted based on high academic performance, the
admission system has failed due to cheating, bribery for admission, exam scores
manipulation and most of the competent candidates are denied admission (Moja,
2000), although the admission policy provides equitable admission into the
universities.
This
situation has affected the standard of students being admitted into tertiary
institutions, students admitted find it difficult to pass their first year
courses, and those who passes do have poor grades, greater number of university
graduates in Nigeria graduates with grades below the second class upper
division and the number of student spilling over is increasing regularly and a
worse scenario is the case in which graduates from tertiary institutions are
not productive in the labor market due to the fact that they perform below
expectation of the employers (Ajaja
2010).
The
inability of the university admission system to give admission to candidate who
will likely do well and other factors has being held responsible for the
decline in the performance of undergraduate student. Hence this study takes a
different approach to solve the problem of admission by exploring how to use an
artificial neural network model to predict the performance of a candidate
before offering the candidate admission. An artificial neural network has the
ability to learn and extrapolate patterns therefore predicting student
performance will be based on extrapolating patterns from historical data of
previous student and their respective performance.
1.1 Statement of the Problem
The
poor quality of graduates of most Nigerian universities is overwhelming and the
ability to predict or forecast the performance of students remains significant
to the growth and development of an institution and the country at large.
The
quality of candidates who are to be admitted into the universities of higher
learning affects the level of training and research within the academic
institution and generally affects the development and growth of the country.
The
inability of the university admission system to give admission to candidate who
will likely do well and other factors has being held responsible for the
decline in the performance of undergraduate student, for this reason this
research work tries to proffer solution by providing a means to evaluate
prospective student who are to be considered for admission, so thereby
admitting those who are likely to perform well in school.
1.2 Aim and Objectives
The
aim of this project work is to use an artificial neural network model to
predict prospective candidate’s performance when admitted. This aim is
achievable through the following objectives
1. To
determine key factors that directly or indirectly affects the performance of
students
2. To
transform highlighted key factors into a form that can be represented in a
neural network
3. To
use transformed factors as background data to train, validate and test an
artificial neural network that can predict a student performance seeking
admission into the university or institution of higher learning.
1.3 Scope of the Study
This
study attempts to identify various factors that affect students’ performance or
factors that have the potential of determining how a candidate will perform
when admitted into the university and uses this factors as background data for
system coding so as to use a suitable artificial neural network model to
predict a prospective student performance.
This
study spans the department of Computer Science, Federal University of
Technology Minna.
1.4 Limitation of the Study
Data
used to train, validate and test the network was obtained from the department
of computer science Federal University of Technology Minna, therefore it may
not be generalize to other department and schools of higher learning.
1.5 Significance of the Study
The
ability to predict or forecast the performance of a prospective candidate
seeking admission will eliminate the problem faced by the university admission
system in determining which student will do well when admitted into the
institution hence improves the University admission system.