CHAPTER ONE
1.1 BACKGROUND TO THE STUDY
The current flow and growth in technology as given rise to increase in the use Data mining (DM) which has gained attention of researchers in the technology industry, businesses and everyday people, this is as a result of large amount of available data and the necessity to turn such data into useful information that can help boost businesses and increase the quality of life (Parneet, Manpreet, & Gurpreet , 2015). DM also known as Knowledge discovery in databases (KDD) involves a process of turning raw data into information that are useful; KDD is a field that has to do with discovering new and possibly useful information in large databases (Abdulsalam, 2016). Interestingly, in recent years there has been major development of data mining technologies to combat challenges associated with KDD by means of mining intelligently in different large databases. Since “knowledge is power” DM focus is mainly on getting implicit, previously unknown useful information from data (Gorunescu, 2011).
A lot of situations in recent years has contributed to the growth in DM as it contributes to computerization of business, it aids in dealings at the government level for various transactions, individuals use intelligent devices that make use of wide range of data and also the prevalent use of World Wide Web as a global information system has swamped us with a large amount of data and information (Jun Peng, Peng, & Ci Fang, 2014).
The use of DM in the Educational field is referred to as Educational Data Mining (EDM); International Educational Data Mining Society described it as “new coming discipline, that has to do with advancement of methods for exploring the exclusive types of data from the Education settings (Dagim, Pro.seema, & Pro.proova, 2018).
Simply put, (EDM) is the application of Data Mining techniques on educational data. EDM solely focuses on analyzing educational data to proffer solutions to educational research issues. It has to do with developing new ways of exploring the educational data and using DM methods to further get insight into student learning environment (Cristobal & Ventura, 2010). EDM gathers raw data from educational sectors and convert them into useful information that has the potential of having massive impact on the education research (Parneet, et al., 2015). Researchers in EDM study wide on various areas including group and personal learning from educational software, computer supported collaborative learning, computer-adaptive testing and also look into issues related with failure of students or lack of interest of some students in relevant courses (Merceron & Yacef, 2003).
In addition, other areas of concentration for EDM include improving model of student; application of EDM methods has been in learning or refining models of domain knowledge structure and studying how to support the education system both in learning software, and other domains, like collaborative learning behaviours (Tang & McCalla, 2005). Basically, there is a rapid increase research focus in the use of DM in educational field in recent years and are majorly concerned with building methods that discover knowledge from originating data from educational environment. This uses different DM techniques such as Decision Trees, Neural Networks, Naïve Bayes, k-Nearest Neighbor and a whole lot more (Parneet, et al., 2015).
It would not be wrong to infer that children are different from one another in various ways like physically, intellectually, academically or intelligently, emotionally and culturally. When a child is slow in learning academically or low in achieving academic skills, they are said to be slow learners (SL) (Priyamvada & Subodh, 2017). SLs are termed to be lazy or clumsy, they do not only lag behind in the classroom but also are affected in other areas such as emotional and psychological well-being. There is an estimate of 5 to 15 percent of children that falls into this category for any given academic setup (Haneesh, Krishnakumar, Sowmya, & Riyaz, 2013).
Out of the various focus of EDM, detecting slow learners and been able to help them in the classroom remains a major area of interest. About 14% of children in the classroom between the age of 4 to 12 are slow learners (Sow, 2017); our educational system does not cater considerably well for them due to the curriculum and system of education followed. These once lag behind and it has adverse effect on their overall well-been as self is affected and it degenerates to a low self-esteem and lack of believe in themselves, it transcends to their family and may induce behavioral defects that affects the society as a whole if care is not taken. Early detection of students that might be at risk for educational failure is a significant process that deserves researcher’s attention. Some students are identified as having special needs when they are at infancy stage but the majority of children are not identified until they enter the school system (Haneesh, et al., 2013).
The challenges faced by slow learners as regards learning in elementary institutions of learning are outlined below:
- Slow learners are usually are usually difficult to detect until after two years of their academic calendar
- Slow learners are often misunderstood by their teachers and are rated as dull, lazy or inept.
- Mixing slow learners with normal children will affect both the teaching staff and the affected children pupils negatively
- Slow learners who have been mixed with normal children often have low self-esteem when it come academic learning which can affect other aspects of their existence and last the rest of their lives
- Slow learners who have been detected late will have difficulty switching to the new learning methods
Hence it is imperative to develop a system that solves this problem early.
Researchers have done a lot on the detection, prediction and a recommender model for the slow learners;
(JothiLaskshmi & Thangaraj, 2019) came up with a Recommender System for Stimulating the Learning Skill of Slow Learner in Higher Educational Institution using EDM, they made use of classification algorithm for learning pattern discovery and the result was used to build a model that recommends learning skills that will stimulate slow learners learning ability.
This study is about the early detection of slow learners in academic setups with the use of appropriate data Mining algorithm and using the result to build a system that will help schools categorize the students into classrooms fit for their learning rate.
1.1.Statement of the Problem
A proper educational system that puts into consideration the slow learners can be achieved with the introduction of EDM, the conventional way has been exploited in dealing with this singular issue; Montessori system has been introduced into the system; this system of education looks at various ways in which children learn individually by watching them for a length of time, noting their interests and what they tend to spend most time doing but this is still limiting in this era of technology surge. EDM has been used to recognize slow learners in a classroom by analyzing the students’ performance using Naive Bayes, SMO, J48, REPTree and Multilayer Perceptron and comparing the output for an accurate result. A research was also done to discover the optimal pattern of learning for elementary school slow learner students through applying integration of two machine learning clustering algorithms Expectation maximization and k-Means.
From review of literature done so far, a lot of work has been done in this area in the higher educational section and little on the elementary sector; however, a gap still exists in the early detection of slow learners in order to help in reducing the academic problems that plague our higher institution, and also to a great extent solve the problems listed above for slow learners. Hence, this research will focus on using developing a slow learner detection and recommender system to detect and stimulate slow learner learning ability.
1.2. Aims and Objectives
The aim of this research is to develop a Slow Learner Detection System using Education Data Mining (SLD-EDM)
The specific objectives ae to:
- to propose a slow learner detection model using Education data mining,
- design a slow learner detection system based on model in (1),
- implement and evaluation the system in (2).
1.3.Methodology Review
In line with the outlined objectives, overall methodology includes;
- A comprehensive literature review will be done on related literature articles through Google scholar over the last 10 years. With specific focus on articles on existing systems for detection of slow learners and factors that are could to have great influence on the performance of a students are identified. In depth study will be done on different Classification algorithms that have been used previously to tackle the problem of slow learners. These algorithms include K-Means, K-Nearest neighbor, Multiple Regression, Navie Bayes, SMO, J48 and others. The weaknesses and strengths of each algorithm will be taken into account as the best algorithm will be chosen for this research work and a model for detection of slow learners will be proposed.
- With the findings on the literature review from (i) static information like academic performance, gender, age, learning preference and class performance will be collected and the information will be converted to *.CSV format for preprocessing and then converted to ARFF format. A total of 200 data will be collected, these data will be analyzed based on students’ performance and the data will be classified into active leaners, average learners and slow learners using.
- The implementation of this system will be done using WEKA 3.8.4 tool for data preprocessing and statistical analysis of different machine learning algorithms. The preprocessed data is inputted into 6 different classification algorithms which are REPTree, Naïve bayes, kNN, SVM, SMO, J48 algorithms for statistical output. The results of various algorithms will be evaluated using precision, recall and f-measure.
1.4.Significance of the study
The research will be helpful in the early detection of slow learners in our classrooms ensuring that proper attention to be given to them early. It will ensure that slow learners will be able to be identified and given the adequate training suitable for their level. This way, they will be able to learn to read and understand without having to feel left out as they would be if joined with normal children. Slow learners will have better overview of themselves and lack of self-confidence that goes along with slow learning will be reduced. It will lead to efficiency in teaching styles of teachers and also increase their productivity and reduce their stress. It will have an overall advantage on the educational system of the society.
1.5.Scope of the study
This study is a build-up on an already existing system for detection of slow learners. All program and experiment will be run on WEKA.
1.6.Definition of Terms
Knowledge Discovery Database (KDD) – Knowledge discovery in databases (KDD) involves a process of turning raw data into information that are useful; KDD is a field that has to do with discovering new and possibly useful information in large databases (Abdulsalam, 2016)
Slow Learner (SL) – Used in describing the condition where a child is slow in learning academically or low in achieving academic skills (Priyamvada & Subodh, 2017).
Education Data Mining (EDM) – This is the use of Data Mining in education field (Dagim, Pro.seema, & Pro.proova, 2018).
SMO Sequential minimal optimization
J48 Decision Tree Algorithm
REPTree Reduced Error Pruning Decision Tree
WEKA Waikato Environment for Knowledge Analysis, developed at the University of Waikato, New Zealand. It is free software licensed under the GNU General Public License, and the companion software to the book “Data Mining: Practical Machine Learning Tools and Techniques
1.8 Organization of Dissertation
This dissertation is organized in five chapters sequentially from chapter one to five.
Chapter one presents the introduction, problem statement, aims and objectives, methodology overview, scope and significance of study.
Chapter two will be made up of literature review or research in slow learners and data mining techniques employed in the detection of slow learners and the review of related studies.
Chapter three will give the description of the methodology that will be employed to achieve the specified objectives.
Chapter four will present the results, evaluation and discussion.
Chapter five will focus on the conclusion and recommendations.
