CHAPTER ONE

1.0 INTRODUCTION

The evolution of Artificial Intelligence (AI) and its successful deployment in different areas of

human endevours have shown that machines have the ability to perform some human-related

tasks including some activities that involves high cognitions. The success stories associated with

the deployment of AI has increased the motivation of scientists to seek new ways or models to

manage data, improve the accuracy of decisions on data, and to improve the quality and access to

education globally. Machine Learning (ML) is one of the emerging technologies from the

artificial intelligence family that are being integrated into the teaching and learning process

(Onyema, 2019). Machine learning is a technique that allows machines to mimic humans both in

terms of reasoning and behaviour (Krish, 2013). The ML technology is taking the performance

of AI systems to the next level due to its features and capabilities. Machine learning present

models that can be trained with data, and then it becomes autonomous over time to understand

the patterns of the data, and is also able to make decisions on the data to solve a given problems

with speed.

Knowledge of machine learning enhance the creation and application of quality, economical and

smart educational materials (Michael Gr. & Abdel-Badeeh, 2020). ML has increased the

academic and scientific debates on the prospective role of Artificial Intelligence (AI) on human

education. Machine learning and other related emerging technologies has the potential to play a

key role in redefining the teaching profession particularly as it relates to the role of a teacher, the

content to be taught; the method of teaching; and the mode of student evaluation. Machine

learning algorithms such as KNearest Neighbour (KNN), Naïve Bayes, Random Forest, Decision

tree and many others have prognostic capabilities that are of great interest to educators. For

1
instance, the learning patterns of students and their performances can be predicted using ML

algorithms. By determining the learning style of students using machine learning, teachers can

easily understand the individual learning differences of their students and then adopt appropriate

measures to diagnose student’s learning difficulties (Elhaj et al., 2020). In this era of big data,

machine learning skills equip students and educators with contemporary knowledge to compete

in modern world (Vijaya & Priya, 2018; Onyema et al., 2020). Educational curriculums have to

be flexible enough to accommodate the teaching and learning of emerging technologies such as

ML in line with 21st century trends in education (Onyema et al., 2022). This would not only

afford students the opportunity to acquire knowledge of recent technologies, but also challenge

the teachers to update their professional knowledge and skills on emerging technologies and

current topics relating to their fields. The inclusion of machine learning in the school curriculum

at the secondary or tertiary level of education are key to the achievement of modern educational

goals that encourages the teaching and learning of modern skills to prepare the learners for the

future.

Computer Science education provides foundations for students to learn digital literacy skills

including recent developments in computing world. Topics in Computer Science evolve over

time due to the constant changes in the world of computing which demands continuous review in

the course curriculum in line with the emerging trends in computer education. The teaching of

Computer Science or ICT cut across all levels of education. Even law students are required to

take “Introduction to Computer Science” as part of the mandatory requirements for completion

of first degree programme in Nigeria. The reason for the inclusion of Computer science as a

General Study (GNS) course for most programmes at the undergraduate level in Nigeria is not

farfetched because of the growing influence of computer technology across all disciplines, which

2
now requires all professionals to possess at least basic knowledge of computer or digital literacy

skills in order to fit into the digital world of works.

1.1 CONCEPT OF MACHINE LEARNING

Machine Learning is part of Computer sciences that evolved from the study of pattern

recognition and computational learning theory in artificial intelligence (Britanian.com, 2020).

According to Arthur Samuel cited in Phil (2013), machine learning occurred when computers are

exposed to training and learning to enhance their abilities to learn and act with or without much

supervision. Machine learning is an automated detection of meaningful patterns in data, and one

of the fastest growing areas of Computer Science, with far-reaching applications (Shai & Shai,

2014). Machine learning is a technique that allows systems to acquire, and integrate knowledge

through large-scale observations, and to improve, and extend itself by learning new knowledge

rather than by being programmed with that knowledge (Beverly, 2009). The workability of

machine learning depends on its ability to understand different features or attributes of a training

datasets or test data which enables it to predict future behaviours or imitate human brains.

According to Krish (2013), machine learning is based on a set of algorithms that can process a

wide variety of data that normally is difficult to process by hand. These algorithms include

Decision tree learning, Neural networks, Naive Bayes, clustering algorithm, K-Nearest Neigbour

(KNN), Genetic algorithm, learning algorithms, explanation-based learning, instance-based

learning, Reinforcement-based learning, Support vector machines, Associative rules,

Recommender algorithms (Krish, 2013).

Machine Learning is a promising technology that can change the narratives in the educational

sector in no distant future both in terms of content, methodology, and evaluation pattern. In

machine learning, machines (computers) are programmed to learn, retain and act independently

3
to solve problems with little or no human interference by manipulating data that has been

provided to the model. Machine learning algorithms learn from data rather than through explicit

programming, and as the algorithms ingest training data, it is then possible to produce more

precise models based on that data (Judith & Daniel, 2018). The machine learning algorithm

update answers to problems as modifications are made to the model. Some machine learning

models are online and continuously adapt as new data is ingested while the other models, called

offline machine learning models, are derived from machine learning algorithms but, once

deployed, do not change (Judith & Daniel, 2018). The digitalization of images, videos, voice and

text has created an environment where machine learning can thrive, and this has allowed AI

researchers to revisit old artificial neural network models, training them with very large datasets

(Tuomi, 2018).

There is an emerging appetite for machine learning approaches as a result of the emergence of

Big Data, and a shift towards more complex instructional domains (Cristina et al., 2018). The

evolution of machine learning has increased the ability of machines to understand and

manipulate complex data and then make decisions on it, but it is still black boxes for many

scholars because of its novelty (Himabindu et al., 2015). Kucak et al. (2018) opined that

Machine learning application in education, would enhance digital learning in the nearest future.

For Ibtehal (2018) the use of Machine Learning and educational technology enabled teachers to

monitor the learning progress of their students and to help special students overcome learning

difficulties. Fedor & Anders (2018) developed a machine learning algorithm with capabilities to

detect students’ prior knowledge of a subject based on symbolic regression, and averred that ML

was an effective algorithm in determining students’ performances. The use of K-NN algorithm is

effective in evaluation of students’ learning styles and had the potential to improve teachers’

4
ability to understand individual student’s learning pattern and future performances (Elhaj et al.,

2020).

According to Steven and Gail (2011) machine learning is an important educational tool which

ease the understanding of concepts by high school students. Therefore, a platform to enhance the

awareness of machine learning technology among schools was developed by them. Xiaojin

(2015) stated that machine learning techniques could be deployed to improve the quality of

training and practicality in education. Trained models can be used to simulate behaviours and

ensure minimal errors by trainees. Himabindu et al. (2015) examined machine learning

framework to identify vulnerable students in classes. They made use of classification algorithms

in line with performance metrics to monitor performances, and assist students with learning

difficulties. Tuomi (2018) assessed different applications of AI in the education sector,

particularly in the classroom. He concluded that AI supports modern teaching and learning

techniques, but it requires proper training of staff to integrate them in the classroom. Popenici &

Kerr (2017), on the impact of AI in teaching and learning in higher education. Opined that many

tasks in education were being replaced by AI-machine learning techniques, and that higher

education institutions should embrace the vast opportunities offered by AI to enrich the quality

of teaching and learning.

Machine learning can be applied in different areas, including pattern recognitionvoice and facial

recognition, biometrics, Sports prediction, driverless vehicles, Filtering system, categorization of

DNA, Games, Customer service, Forecasting, searching, Ranking and recommending and

prototyping. The predictive ability of machine learning algorithms is helping many companies

and organizations to predict customers’ or users’ behaviours based on their activities on the

platform. For instance, big players in the technology industry such as Amazon, Google, Apple,

5
Facebook, Twitter, and Netflix are using machine learning algorithms to improve user

experiences by speeding and matching up their search. With the aid of machine learning

capabilities, the personality of a user can be accurately predicted based on his or her information

available on the database and the previous activities on the account or page. Machine learning

are being deployed on social networks to enhance user experiences, including the ability to relate

images, relationships and locations. Posts can easily be tracked and filtered for possible

community abuse. Judith & Daniel (2018) states that machine learning technology is embedded

in applications throughout enterprises in order to improve performance, increase customer

satisfaction, reduce customer churn, and boost revenue. Machine Learning enhances the

understanding of data and the identification of patterns of data which can assist institutions and

organizations to improve the quality of their products and customer relations.

The prediction of users’ action provides guide for vendors to recommend products and services

to potential customers exposing them to wide range of options and beguiling them to buy more.

Vishal & Sonawane (2016) stated that machine learning techniques like; the Naive Bayes (NB),

Maximum Entropy (ME), and Support Vector Machines (SVMs) have achieved great success in

sentiment analysis. There is an increasing number of educational institutions that are fast

embracing machine learning techniques to improve their efficiencies and speed (Ibtehal, 2018).

One of the benefits of incorporating machine learning techniques in education is the automation

aspect of enriching the knowledge base with self-learning techniques with minimal human

intervention in the process (Krish, 2013).

6
 Figure 1: Machine Learning Process (Krish, 2013).

Consequently, Judith and Daniel (2018), categorized the training of a machine learning algorithm

to create an accurate model into three steps:

Representation: The algorithm creates a model to transform the inputted data into the desired

results. As the learning algorithm is exposed to more data, it will begin to learn the relationship

between the raw data and which data points are strong predictors for the desired outcome (Judith

& Daniel, 2018; Elhaj et al., 2020).

i. Evaluation: As the algorithm creates multiple models, either a human or the algorithm

will need to evaluate and score the models based on which model produces the most

accurate predictions. It is important to remember that after the model is operationalized, it

will be exposed to unknown data. As a result, make sure the model is generalized and not

overfit to your training data.

ii. Optimization: After the algorithm creates and scores multiple models, select the best

performing algorithm. As you expose the algorithm to more diverse sets of input data,

select the most generalized model (Judith & Daniel, 2018).

1.2 CATEGORIZATION OF MACHINE

7
The categorization of Machine learning technique is based on the type of datasets. Tutorials

point (2019) categorized machine learning into supervised learning; unsupervised learning;

reinforcement learning; deep learning; and deep reinforcement learning as shown in figure 2.

Figure 2: Categorization of Machine Learning (Tutorials point, 2019).

Supervised learning takes place if the model is fed with data that has labels while the

Unsupervised learning deals with unlabeled data. Both supervised and unsupervised learning

have their pros and cons depending on the use of the model.

Supervised Learning

Supervised learning uses historical data to predict future actions or events. This means that its

function relies heavily on the past without much consideration of likely changes or innovations

which can alter events. The supervised learning algorithms use a known dataset (called the

training dataset) to make predictions. Some of the algorithms for supervised learning include:

Naives Bayes, K-Nearest Neigbour Algorithm, Support Vector machines, Decision trees and

Logistic regression.

Unsupervised Learning

8
Unsupervised Learning also known as self-organizing maps is a type of machine learning that are

used to group cases based on similar attributes, or naturally occurring trends, patterns, or

relationships in the data (Colleen, 2015). The unsupervised learning algorithms are type of

machine learning algorithms used for inferences from datasets consisting of input data without

labeled responses (Elhaj et al., 2020), and it is based on rewards and punishments. Unsupervised

learning aims at extracting regularities from datasets, in order to simplify their description by

reducing them to their most characteristic elements (Lesot, 2006). It can be applied in facial

recognition technologies and others.

Reinforcement Learning

Reinforcement learning is an aspect of machine learning that deals with how software agents

ought to take actions in an environment in order to maximize the notion of cumulative reward

(Wikipedia, 2020). Reinforcement learning deals with the ability of learning the associations

between stimuli, actions, and the occurrence of pleasant events, called rewards, or unpleasant

events called punishments (Stefano & Mathias, 2013). It is a type of machine learning technique

that enables an agent to learn in an interactive environment by trial and error using feedback

from its own actions and experiences (Kdnuggets.com, 2018). It can be applied to different areas,

including Operation Research due to its generality and usefulness.

Deep Learning

This is a kind of machine learning that enhances the understanding and interpretation of data,

features and knowledge. It involves the imitation of human brain, giving the model the ability to

think and act like humans. Deep learning technique is used for extracting high-level abstract

features, providing improved performance over the traditional models, increasing interpretability

9
and also for understanding and processing of biological data (Navamani, 2019). Deep learning

models such as Deep Neural Networks (DNN) and Convolution Neural Networks (CNN) are

being used for categorizing reactions, cross-lingual problems as well as product review analysis

(Koyel et al., 2019).

1.3 THE AIM OF MACHINE LEARNING

The field of machine learning can be organized around three primary research Areas:

i. Task-Oriented Studies: The development and analysis of learning systems oriented

toward solving a predetermined set, of tasks (also known as the “engineering approach”).

ii. Cognitive Simulation: The investigation and computer simulation of human learning

processes (also known as the “cognitive modelling approach”)

iii. Theoretical Analysis: the theoretical exploration of the space of possible learning

methods and algorithms independent application domain.

Although many research efforts strive primarily towards one of these objectives, progress in on

objective often lends to progress in another. For example, in order to investigate the space of

possible learning methods, a reasonable starting point may be to consider the only known

example of robust learning behaviour, namely humans (and perhaps other biological systems)

Similarly, psychological investigations of human learning may held by theoretical analysis that

may suggest various possible learning models. The need to acquire a particular form of

knowledge in stone task-oriented study may itself spawn new theoretical analysis or pose the

question: “how do humans acquire this specific skill (or knowledge)?” The existence of these

mutually supportive objectives reflects the entire field of artificial intelligence where expert

10
system research, cognitive simulation, and theoretical studies provide some (cross-fertilization of

problems and ideas (Jaime G. Carbonell, 1983).

1.3.1 Applied Learning Systems

At, present, instructing a computer or a computer-controlled robot, to perform a task requires one

to define a complete and correct, algorithm for that task, and then laboriously program the

algorithm into a computer. These activities typically involve a tedious and time-consuming effort

by specially trained personnel. Present-day computer systems cannot truly learn to perform a task

through examples or by analogy to a similar, previous-solved task. Nor can they improve

significantly on the basis of past, mistakes or acquire new abilities by observing and imitating

experts. Machine learning research strives to open the possibility of instructing computers in

such new ways, and thereby promises to ease the burden of hand-programming growing volumes

of increasingly complex information into the computers of tomorrow. The rapid expansion of

application and availability of computers today makes this possibility even more attractive and

desirable.

1.3.2 Knowledge Acquisition

When approaching a task-oriented knowledge acquisition task, one must be aware that, the

resultant computer system must interact with humans, and therefore should closely parallel

human abilities. The traditional argument that an engineering approach need not reflect human or

biological performance and is not, truly applicable to machine learning. Since airplane, a

successful result on an almost pure engineering approach, better little resemblance to their

biological counterparts, one may argue that applied knowledge acquisition systems could be

equally divorced from any consideration of human capabilities. This argument does not apply

11
here because airplanes need not interact, with or understand birds Learning machines, on the

other hand, will have to interact, with the people who make use of them, and consequently the

concept and skills they acquire- if not necessarily their internal mechanism and must be

understandable to human.

1.4 MACHINE LEARNING AS A SCIENCE

The clear contender for a cognitive invariant in human is the learning mechanism which is the

ability facts, skills and more abstractive concepts. Therefore understanding human learning well

enough to reproduce aspect of that learning behaviour in a computer system is, in itself, a worthy

scientific goal. Moreover, the computer can render substantial assistance to cognitive

psychology, in that it may be used to test the consistency and completeness of learning theories

and enforce a commitment to the fine-structure process level detail that precludes meaningless

tautological or untestable theories (Bishop, 2006). The study of human learning processes is also

of considerable practical significance. Gaining insights into the principles underlying human

learning abilities is likely to lead to more effective educational techniques. Machine learning

research is all about developing intelligent computer assistant or a computer tutoring systems and

many of these goals are shared within the machine learning fields. According to Jaime et al

(Jaime G. Carbonell, 1983) who stated computer tutoring are starting to incorporate abilities to

infer models of student competence from observed performance. Inferring the scope of a

student’s knowledge and skills in a particular area allows much more effective and

individualized tutoring of the student (Sleeman, 1983).

12
The basic scientific objective of machine learning is the exploration of possible learning

mechanisms, including the discovery of different induction algorithms, the scope of theoretical

limitations of certain method seems to be the information that must be available to the learner,

the issue of coping with imperfect training data and the creation of general techniques applicable

in many task domains. There is no reason to believe that human learning methods are the only

possible mean of acquiring knowledge and skills. In fact, common sense suggests that human

learning represents just one point in an uncharted space of possible learning methods- a point

that through the evolutionary process is particularly well suited to cope with the general physical

environment in which we exist. Most theoretical work in machine learning are centred on

creation, characterization and analysis of general learning methods, with the major emphasis on

analyzing generality and performance rather than psychological plausibility. Whereas theoretical

analysis provides a means of exploring the space of possible learning methods, the task-oriented

approach provides a vehicle to test and improve the performance of functional learning systems

and testing applied learning systems, one can determine the cost-effectiveness trade-offs and

limitations of particular approaches to learning. In this way, individual data points in the space

possible learning systems are explored and the space itself becomes better understood.

Knowledge Acquisition and Skill Refinement: There are two basic form of learning:

i. Knowledge Acquisition

ii. Skill refinement

When it is said that someone learned mathematics, it means that this person acquired concepts of

mathematics, understood the meaning and their relationship to each other as well as to the world.

The importance of learning in this case is acquisition of knowledge, including the description

and models of physical systems and their behaviours, incorporating a variety of representations

13
from simple intrusive mental model models, examples and images to completely test

mathematical equations and physical laws. A person is said to have learned more if this

knowledge explains a broader scope of situations, is more accurate, and is better able to predict

the behaviour of the typical world (Allix, 2003). This form of learning is typically to a large

variety of situations and is generally learned knowledge acquisition. Therefore, knowledge

acquisition is defined as learning a new task coupled with the ability to apply the information in

the effective manner.

The second form of learning is the gradual improvement of motor and cognitive skills through

practice- Learning by practice. Learning such as:

 Learning to drive a car

 Learning to play keyboard

 Learning to ride a bicycle

 Learning to play piano

1.5 MACHINE LEARNING APPLICATIONS

The other aspect for classifying learning systems is the area of application which gives a new

dimension for machine learning. Below are areas to which various existing learning systems

have been applied. They are:

i. Computer Programming

ii. Game playing (chess, poker, and so on)

iii. Image recognition, Speech recognition

iv. Medical diagnosis

v. Agriculture, Physics

14
 vi. Email management, Robotics

vii. Music

viii. Mathematics

ix. Natural Language Processing and many more.

CHAPTER TWO

2.0 LITERATURE REVIEW

2.1 INTRODUCTION

Some studies, e.g., Vinicius H. S. et al., 2019, Santiago Fajardo, et al., 2016, have already

analyzed the application of ML in SE in the past. Durelli et al., conducts a systematic mapping

study on the application of ML for software testing. The study highlights the use of ML

techniques in various software testing activities such as testcase generation and oracle

construction. Results of the study show that a vast majority of articles employ supervised

learning, such as ANN and DT, to solve testing-related problems. Moreover, the key advantages

and disadvantages of using ML for software testing are discussed. Mainly, the advantage of ML

techniques is their scalability and efficient application to large-scale and complex software

systems. The disadvantage, on the other hand, is the unavailability of data that fits well with the

learning process.

Fajardo et al., provides an extensive overview of applying data mining techniques to SE tasks

including open issues and recommendations for improvements. The study provides a

15
comprehensive list of data mining techniques applicable in SE, e.g., aspects related to clustering,

regression, and performance metrics. Moreover, the study highlights some widely used datasets

of SE employed in the data mining articles and states key advantages of mining SE data.

Wan et al., performed a survey by interviewing 14 people from 3 companies and 342

respondents from 26 countries. The aim of the study was to understand and highlight the key

differences in the software development practices followed in building ML and non-ML software

systems. Results suggested that ML engineers should focus on handling the uncertain

randomness of ML systems and work on employing version control systems specifically for ML

applications in order to improve reproducibility.

Zhang et al., conducted the research focusing on the application of ML in SE. The study provides

a comprehensive list of SE tasks whose problems can be addressed using ML techniques. The

study also emphasizes the fact of SE to be a highly fertile area to explore with regards to

applying ML techniques by formulating SE tasks as learning problems and addressed using ML

techniques.

In contrast to the aforementioned focused studies, our study provides a broader context and a

comprehensive list of articles that help identify the relationship between various ML techniques

and SDLC stages. We also provide the relationships of ML types, tools, and techniques with

respect to SE stages, which help better understanding the current progress of the adoption of ML

for SE.

2.2 CATEGORISATION OF MACHINE LEARNING ALGORITHMS

An overwhelming number of ML algorithm have been designed and introduced over past years.

Not every one of them are widely known. Some of them did not satisfy or solve the problem, so

16
another was introduced in its place. Here the algorithms are broadly grouped into two category

and those two groups are further sub-divided. This section try to name most popular ML

algorithms and the next section compares three most widely used ML algorithms.

A. Algorithms Grouped By Similarity

1. Regression Algorithms: Regression analysis is part of predictive analytics and exploits the

co-relation between dependent (target) and independent variables. The notable regression models

are: Linear Regression, Logistic Regression, Stepwise Regression, Ordinary Least Squares

Regression (OLSR), Multivariate Adaptive Regression Splines (MARS), Locally Estimated

Scatterplot Smoothing (LOESS) etc.

2. Instance-based Algorithms: Instance-based or memory-based learning model stores instances

of training data instead of developing a precise definition of target function. Whenever a new

problem or example is encountered, it is examined in accordance with the stored instances in

order to determine or predict the target function value.It can simply replace a stored instance by a

new one if that is a better fit than the former. Due to this, they are also known as winner-take-all

method. Examples: K-Nearest Neighbour (KNN), Learning Vector Quantisation (LVQ), Self-

Organising Map (SOM), Locally Weighted Learning (LWL) etc.

3. Regularization Algorithm: Regularization is simply the process of counteracting over fitting

or abate the outliers. Regularization is just a simple yet powerful modification that is augmented

with other existing ML models typically Regressive Models. It smoothes up the regression line

by castigating any bent of the curve that tries to match the outliers. Examples: Ridge Regression,

Least Absolute Shrinkage and Selection Operator (LASSO), Elastic Net, Least-Angle Regression

(LARS) etc.

17
4. Decision Tree Algorithms: A decision tree constructs a tree like structure involving of

possible solutions to a problem based on certain constraints. It is so named for it begins with a

single simple decision or root, which then forks off into a number of branches until a decision or

prediction is made, forming a tree. They are favoured for its ability to formalize the problem in

hand process that in turn helps identifying potential solutions faster and more accurately than

others. Examples: Classification and Regression Tree (CART), Iterative Dichotomiser 3 (ID3),

C4.5 and C5.0, Chi-squared Automatic Interaction Detection (CHAID) , Decision Stump, M5,

Conditional Decision Trees etc.

5. Bayesian Algorithms: A group of ML algorithms employ Bayes’ Theorem to solve

classification and regression problems. Examples: Naive Bayes, Gaussian Naive Bayes,

Multinomial Naive Bayes, Averaged One-Dependence Estimators (AODE), Bayesian Belief

Network (BBN), Bayesian Network (BN) etc.

6. Support Vector Machine (SVM): SVM is so popular a ML technique that it can be a group

of its own. It uses a separating hyper plane or a decision plane to demarcate decision boundaries

among a set of data points classified with different labels. It is a strictly supervised classification

algorithm. In other words, the algorithm develops an optimal hyper plane utilizing input data or

training data and this decision plane in turns categories new examples. Based on the kernel in

use, SVM can perform both linear and nonlinear classification.

7. Clustering Algorithms: Clustering is concerned with using ingrained pattern in datasets to

classify and label the data accordingly. Examples: K-Means, K-Medians, Affinity Propagation,

18
Spectral Clustering, Ward hierarchical clustering, Agglomerative clustering. DBSCAN, Gaussian

Mixtures, Birch, Mean Shift, Expectation Maximization (EM) etc.

8. Association Rule Learning Algorithms: Association rules help discover correlation between

apparently unassociated data. They are widely used by e- commerce websites to predict customer

behaviours and future needs to promote certain appealing products to him. Examples: Apriori

algorithm, Eclat algorithm etc.

9. Artificial Neural Network (ANN): Algorithms A model based on the built and operations of

actual neural networks of humans or animals.ANNs are regarded as non-linear modelsas it tries

to discover complex associations between input and output data. But it draws sample from data

rather than considering the entire set and thereby reducing cost and time. Examples: Perceptron,

Back- Propagation, Hop-field Network, Radial Basis Function Network (RBFN) etc.

10. Deep Learning Algorithms: These are more modernised versions of ANNs that capitalise

on the profuse supply of data today. They are utilizes larger neural networks to solve semi-

supervised problems where major portion of abound data is unlabeled or not classified.

Examples: Deep Boltzmann Machine (DBM), Deep Belief Networks (DBN), Convolutional

Neural Network (CNN), Stacked Auto-Encoders etc.

2.3 MEASURING AND COMPARING PERFORMANCES OF POPULAR ML

ALGORITHMS

Though various researchers have contributed to ML and numerous algorithms and techniques

have been introduced as mentioned earlier, if it is closely studied most of the practical ML

approach includes three main supervised algorithm or their variant. These three are namely,

Naive Bayes, Support Vector Machine and Decision Tree. Majority of researchers have utilised

19
the concept of these three, be it directly or with a boosting algorithm to enhance the efficiency

further. These three algorithms are discussed briefly in the following section.

a. Naive Bayes Classifier

It is a supervised classification method developed using Bayes’ Theorem of conditional

probability with a ‘Naive’ assumption that every pair of feature is mutually independent. That is,

in simpler words, presence of a feature is not effected by presence of another by any means.

Irrespective of this over-simplified assumption, NB classifiers performed quite well in many

practical situations, like in text classification and spam detection. Only a small amount of

training data is need to estimate certain parameters. Beside, NB classifiers have considerably

outperformed even highly advanced classification techniques.

b. Support Vector Machine

SVM, another supervised classification algorithm proposed by Vapnik in 1960s have recently

attracted a major attention of researchers. The simple geometrical explanation of this approach

involves determining an optimal separating plane or hyper plane that separates the two classes or

clusters of data points justly and is equidistant from both of them. SVM was defined at first for

linear distribution of data points. Later, the kernel function was introduced to tackle nonlinear

datas as well.

c. Decision Tree

A classification tree, popularly known as decision tree is one of the most successful supervised

learning algorithm. It constructs a graph or tree that employs branching technique to demonstrate

every probableresult of a decision. In a decision tree representation, every internal node tests a

feature, each branch corresponds to outcome of the parent node and every leaf finally assigns the

20
class label. To classify an instance, a top-down approach is applied starting at the root of the tree.

For a certain feature or node, the branchconcurring to the value of the data point for that attribute

is considered till a leaf is reached or a label is decided.

CHAPTER THREE

DISCUSSION

3.1 PROSPECTS OF MACHINE LEARNING IN COMPUTER SCIENCE

EDUCATION

The rapid technological development of the last decades created new complex problems and

changes in the education sector, the solution of which requires not only critical thinking, but also

another mode of advanced thinking that has been called Computational Thinking (Michael Gr. &

Abdel-Badeeh, 2020). The educational system in the 21st century has gone beyond the wall of

classrooms or the positioning of teachers as masters of knowledge. The contemporary teaching

and learning process requires current knowledge and use of modern technologies to achieve the

set objectives. Machine learning is among the recent technologies that is transforming the

educational sector. It is an aspect of Artificial Intelligence (AI) that embeds human reasoning

and behaviour to machines, giving them the ability to perform human related tasks autonomously

based on pattern threads of activities or data inputted into the machine.

21
Ibtehal (2018) opined that machine learning is a subset of Artificial Intelligence that enable

computers or teaching devices to learn from all previous data and make intelligent decisions.

Xiaojin (2015) states that machine teaching can be used to design optimal lesson for individual

students and to maximally influence the learner via optimal training data. Machine Learning can

help educational institutions to test the workability of a strategy, before adopting it. Gordon

(2019) stated that machine learning is optimizing and personalizing learning experiences for

students and helping teachers. Machine learning can potentially foster quality learning and

redefine how education is delivered (Pallavi & Bramah, 2019).

The growing shortage of qualified teachers coupled with growing numbers of students world-

wide, represents a substantial societal global challenge and provides a strong motivation to

continue to invest in solutions to enhance and support human learning and development (Cristina

et al., 2018). Machine Learning offers some help to educators on learners’ mental and cognitive

ability. Machine learning can also act as a virtual assistant - providing automatic and real-time

responses to queries by users on interactive forums or tutorial platforms. Machine learning

enhances the efficiencies of educators by enabling them to complete multiple tasks such as

classroom management and scheduling at ease (thetechedvocate.org, 2019). The adoption of

Machine learning enhanced the development of free, user-friendly software and other

educational content that is easily understandable to general audience and available over the

internet (Vijaya & Priya, 2018). Machine learning based assessment provides constant feedback

to teachers, students and parents about how the student learns, the support they need and the

progress they are making towards their learning goals (Kucak et al., 2018). Machine learning

presents a lot of opportunities and challenges to computer scientists on how to find answers to

22
numerous questions using mathematical equations and programming codes. Other prospects of

machine learning in education are summarized as follows:

a. Mitigating Examination Malpractices: The ability of machine learning algorithms to

accurately assess students present futuristic solutions to errors and malpractices

associated with human involvement in students’ assessment or grading. The ability of

machine learning to recognize speeches and handwriting can also be useful in

checkmating academic frauds.

b. Improved Students Performance: Prediction of students learning styles or performance

provides teachers with opportunity to put interventions in place to prevent students’

failure. By identifying “at risk” students and weaknesses early, schools can detect and

contact those students and help them to be more successful (Kucak et al., 2018).

c. Academic Translation: Educational books can easily be translated into different

languages with the aid of machine learning algorithms. Machine learning models can also

be applied in testing the reliability and validity of educational software.

d. Teacher Support: Machine learning techniques could be used to perform routine tasks

like class attendance and customized lectures, thereby reducing the workloads of the

teacher. ML can also improve teaching by repeatedly observing how students react and

generalize rules about the domain or student (Beverly, 2009), thereby assisting the

teacher to generate feedbacks and improve their quality of teachings.

e. Plagiarism Detection: Machine learning techniques are been used to develop plagiarism

software which are often used to improve the quality of papers for publication and to

enhance academic integrity.

23
f. Promotion of Learning Interests: Through recommendations and filtering systems,

machine learning solutions often motivate learners to increase their reading skills or

develop interests in a given topic being recommended.

g. School Finance: The use of machine learning solutions enhances security of financial

records or transactions of educational institutions. Machine learning algorithms are useful

in detection of credit card frauds in schools. Similarly, the wave of interest and

investments in artificial intelligence could also improve the revenue of educational

institutions, particularly those who are involved in provision of AI products or solutions

(Popenici & Kerr, 2017). The use of machine learning solutions could reduce cost for

educational institutions in the future.

h. Research: Machine learning has many features that could enhance practicality in

education and research. For instance, researchers often use machine learning technique to

carry out detection, perception and sentiment analysis which are very vital in data mining

and prediction of research outcomes.

i. Security: The implementation of machine learning in education has brought about the

production of AI-driven trackers which can be attached to school buses or school hostels

to support campus security (medium.com, 2020). There are also emerging robots which

are being used for campus security and anti-spam software that help filter academic

emails. ML is also useful in school biometric designs for staff and students.

j. Career or School Choices: The use of machine learning prediction model could help

educational institutions in making subject placement decisions for students based on their

activities or attributes. Students can also objectively analyze their skills, abilities, and

24
 interests, and based on them, pick the right schools or path after graduation

(medium.com, 2020).

k. Automated Grading System: The development of automated paper grading systems is

being made possible through the use of machine learning algorithms. Machine learning is

helping teachers to grade quicker and with a greater accuracy (Gordon, 2019), thereby

reducing bias and human interferences, and also quickening result computation.

3.2 LIMITATIONS OF MACHINE LEARNING

Nearly all technologies have their limitations and machine learning is not an exception. A

computer program is said to learn from experience (E) with respect to some class of tasks (T)

and performance measure (P), if its performance at tasks in T, as measured by P, improves with

experience (Tom, 1997). Consequently, the accuracy of outputs or decisions expected from

machine learning technique largely depends on the quality of experience and input in the

machines. Machine learning techniques are prone to machine mistakes. Errors can emanate from

data mining process, subjectivity, classification, or interpretations etceteras. Machine learning is

also new and might have some flaws that could mar its precision and accuracy, particularly as it

relates to educational grading systems. Popenici and Kerr (2017) warned that education is

eminently a human-centric endeavor, not a technology centric solution, thus, the idea that we can

solely rely on AI or any other technologies is a dangerous path. While machine learning brings

great promise for the future of education, relying on it totally would be a big mistake

(givingcompass.org, 2020). The application of machine learning in the teaching and learning

process can take away personal interactions from the students which could negatively impact

their ability to make friends and present themselves well in the workplace over the years ahead.

(Thetechedvocate, 2018).

25
There is skepticism about security of data, missing data, inaccuracies in predictions, over fitting

of small data, and the fears over possible replacement and displacement of workers by machines

in the nearest future. If machines can work independently and effectively including as a tutor,

many employees across different professions particularly those in the academic world may likely

be displaced from their jobs. A number of researchers have developed machine learning

algorithms to assist educators but the challenge is improving the accuracy, sensitivity, and

specificity of the machine learning algorithms (Fedor & Anders, 2018). Machine learning

algorithms are prone to different errors including classifications errors which can affect its output

and acceptability. Similarly, Popenici and Kerr (2017) highlighted the need for educational

institutions to be mindful of the use of Machine learning /AI solutions to avoid privacy breaches

by those who control the algorithms. Machine learning technique can help streamline and

improve the teaching and learning process, but they cannot replace the cultural element of

learning, which can only come from another human (givingcompass.org, 2020).

26
 CHAPTER FOUR

4.0 CONCLUSION

The significant pedagogical potentials of machine learning technology can be tapped to improve

the standard or quality of teaching and learning, particularly as it relates to problem and

practical-based Computer Science education. This therefore call for the need to observe caution

during its implementation in Computer Science education to avoid overhyping the technology,

and to appreciate the human and cultural elements in education.

27
REFERENCE

Allix, N. M. (2003, April). Epistemology And Knowledge Management Concepts And Practices.

Journal of Knowledge Management Practice .

Bis*hop, C. M. (1995). Neural Networks for Pattern Recognition. Oxford, England: Oxford

University Press.

Bishop, C. M. (2006). Pattern Recognition and Machine Learning (Information Science and

Statistics). New York, New York: Springer Science and Business Media.

Colleen, M (2015). Identification, Characterization, and Modeling. Data Mining and Predictive

Analysis (Second Edition)-Intelligence Gathering and Crime Analysis, 137-155.

https://doi.org/10.1016/B978-0-12-800229-2.00007-9.

28
Cristina, C; Ka´ska, P; and Manolis, M. (2018). AI in Education needs interpretable Machine

Learning: Lessons from Open Learner Modelling. Paper presented at 2018 ICML

Workshop on Human Interpretability in Machine Learning (WHI 2018), Stockholm,

Sweden.

Du Zhang. Machine Learning and Software Engineering. Software Quality Journal, 11(3):87–

119,

2003.

Elhaj, M.A.E; Bashir, S.G; Abdullahi, I; Onyema, E.M; Hauwa, A.A; Hayatu, A.S. (2020).

Evaluation of the Performance of K-Nearest Neighbor Algorithm in Determining Student

Learning Styles. International Journal of Innovative Science, Engineering. &

Technology., 7 (1), 91-102. ISSN(Online) 2348- 7968.

Fedor, D. & Anders,G. (2018). Machine Learning-Based App for Self-Evaluation of Teacher-

Specific Instructional Style and Tools. Educ. Sci. 8 (7),1-15.

doi:10.3390/educsci8010007.

Givingcompass.org Website (October, 2020). The Benefits And The Limitations Of

Machine Learning In Education. Available at:

https://www.givingcompass.org/article/the- benefits-and-the-limitations- ofmachine-

learning-in-education/. Accessed 24 October 2020.

Himabindu, L; Everaldo, A; Carl, S; David,M; Nasir, B; Rayid, G. and Kecia, L.A. (2015). A

29
 Machine Learning Framework to Identify Students at Risk of Adverse Academic

Outcomes.ACM.ISBN978-1-4503-3664-2/15/08,DOI:

http://dx.doi.org/10.1145/2783258.2788620.

I. H.Sarker, "SPRINGER," FRI MARCH 2021. [Online]. Available: Sarker, I.H. Machine

Learning: Algorithms, Real-World Applications and Research Directions. SN COMPUT.

SCI. 2, 160 (2021). https://doi.org/10.1007/s42979-021-00592-x. [Accessed FRI

MARCH 2021].

Ibtehal, T.N. (2018). Machine Learning in Educational Technology. Chapter 9, pp175-

183.Retrieved from IntechOpen via: http://dx.doi.org/10.5772/intechopen.72906.

Judith, H. and Daniel, K. (2018). Machine Learning For Dummies. IBM Limited Edition.

ISBN: 978-1-119-45494-6 (ebk).

Kdnuggets.com website (2018). 5 Things You Need to Know about Reinforcement Learning.

Available at: https://www.kdnuggets.com/2018/03/5-thingsreinforcement-learning.html.

Accessed 24 October 2020.

Kucak, D; Juricic, V & Dambic, G. (2018). Machine Learning in Education - a Survey of

Current

Research Trends. Proceedings of the 29th DAAAM International Symposium, pp.0406-

0410, B. Katalinic (Ed.), Published by DAAAM International, ISBN 978-3-902734-20-4,

ISSN 1726-9679, Vienna, Austria.DOI: 10.2507/29th.daaam.proceedings.059.

Krish, K. (2013). Data-Driven Architecture for Big Data. Data Warehousing in the Age of

30
 BigData.-MK Series on Business Intelligence, 219-240. https://doi.org/10.1016/B978-0-

12-405891-0.00011-8.

Koyel, C; Siddhartha, B; Rajib, B; Aboul, A. (2019). Sentiment Analysis on a Set of Movie

Reviews Using Deep Learning Techniques. Social Network AnalyticsComputational

Research Methods and Techniques, 127-147. https://doi.org/10.1016/B978-0-12-815458-

8.00007-400.

Navamani, T.M. (2019). Efficient Deep Learning Approaches for Health Informatics. Deep

Learning and Parallel Computing Environment for Bioengineering Systems, 123-137.

https://doi.org/10.1016/B978-0-12-816718-2.00014-2.

Onyema, E.M. (2019). Integration of Emerging Technologies in Teaching and Learning Process

in Nigeria: the challenges. Central Asian Journal of Mathematical Theory and Computer

Sciences, 1 (August), 1. 35-39.

Onyema ,E.M; Sharma, A; Nwafor, C.E; Fyneface, A.G; Sen, S; Edeh, E.C. (2020). Impact of

Emerging Technologies on the Job Performance Of Educators in Selected Tertiary

Institutions in Nigeria. The Journal of Computer Science and its Applications, 27 (1), 52-

62. https://dx.doi.org/10.4314/jcsia.v27i1.4

Popenici, S.A.D. & Kerr, S. (2017). Exploring the Impact of Artificial Intelligence on Teaching

and Learning in Higher Edu. Research &Practice in Techn. Enhanced Learning, 12 (22),

1-13. DOI 10.1186/s41039-017-0062-8.

Santiago Fajardo, García-Galvan, Federico R., Violeta Barranco, Juan C. Galvan, and Sebastian

31
 Feliu Batlle. Data Mining and Machine Learning for Software Engineering. Intech,

i(tourism):13, 2016.

Sleeman, D. H. (1983). Inferring Student Models for Intelligent CAI. Machine Learning. Tioga

Press.

Stefano, P. & Mathias, P. (2013). Reinforcement Learning and Tourette Syndrome. International

Review of Neurobiology, 112, 131-153. https://doi.org/10.1016/B978-0-12-411546-

0.00005-6

Shai, S. & Shai, B. (2014). Understanding Machine Learning: From Theory to Algorithms.

Published by Cambridge University Press. ISBN 978-1-107- 05713- 5 Hardback.

Tuomi, I. (2018). The Impact of Artificial Intelligence on Learning, Teaching, and

Education. Policies for the future, Eds. Cabrera, M., Vuorikari, R & Punie, Y., EUR

29442 EN, Publications Office of the European Union, Luxembourg, 2018, ISBN 978-

92-79-97257-7, doi:10.2760/12297, JRC113226.

Tutorialspoint (2019). Machine Learning. Published online by Tutorials Point (I) Pvt. Ltd.

Retrieved online via www.tutorialspoint.com

Vinicius H. S. Durelli, Rafael S. Durelli, Simone S. Borges, Andre T. Endo, Marcelo M. Eler,

Diego R. C. Dias, Marcelo P Guimar, and Marcelo P. Guimaraes. Machine Learning

Applied to Software Testing : A Systematic Mapping Study. IEEE Transactions on

Reliability, pages 1–24, 2019

Vishal,A.K. & Sonawane,S.S.(2016). Sentiment Analysis of Twitter Data: A Survey of

32
 Techniques. Int. Journal of Computer Applications, 139 (11), 5-15.

Vijaya, B. K. & Priya S. G. (2018). Machine Learning and Medical Education. npj Digital

Medicine, 1 (54 ), 1-3. doi:10.1038/s41746-018-0061-1

Zhiyuan Wan, Xin Xia, David Lo, and Gail C. Murphy. How does Machine Learning Change

Software Development Practices? IEEE Transactions on Software Engineering, PP(c):1–1, 2019.

33