1

INTRODUCTION
1.1 Motivation
1.2 Problem statement
1.3 Data mining
1.4 Statistical machine learning
1.5 Learning disability
C o n t e n t s

1.5.1 Concept of learning disability

1.5.2 Basic types of learning disabilities
1.5.3 Assessment and remediation
1.6 Objectives of the present study
1.7 Scope of work
1.8 System framework
1.9 Thesis contributions
1.10 Road map
1.11 Contemporary works
1.12 Summary and conclusion

1.1 Motivation
Learning disability or specific learning disability is a lifelong neuro-
developmental disorder which manifest in childhood as persistent difficulties in
learning to efficiently read, write or do simple mathematical calculations
despite normal intelligence, conventional schooling, intact hearing and vision,
adequate motivation and socio-cultural opportunity [1]. It is now known that a
learning disability is not connected to mental retardation. The learning disabled
frequently have high IQs. It is also not a single disorder, but includes
disabilities in any of areas related to reading, language and mathematics.

During the past few decades, the understanding of learning disability has
changed. However, it is a tremendous challenge to identify and diagnose and
Chapter-1

assist children with learning disability. As the concept is still new, in many
developing countries including India, the research conducted in learning
disability has been primarily done over the last two decades are yet in the
infancy stage. In India, the research conducted in learning disability has been
primarily done over the last two decades and is today comparable with the
research carried out in west nearly half a century ago.

Since no national census of the learning disabled has been taken in India,
it is difficult to collect their actual number. In India, the learning disabled
children are not identified using reliable tests. However, at least 10% of
children in India have a learning disability [2]. We do not have a clear idea
about incidence and prevalence of learning disability in India. These facts
suggest that the early diagnosis of learning disability in children is critically
important to identify and suggest remedial solutions to the parents and children
to understand about the learning disability as stumbling blocks such as lack of
awareness, indifference and apathy and hamper success.

The problems of LD affected children have been a cause of concern to

parents and school authorities for some time. With the right help at right time,
right assessment and remediation, children with LD can and do learn
successfully and become winners in the society later. Since LD has distinctive
symptoms in its early stage, diagnosis approaches have been improved
noticeably over the past decades in many counties. Research works done in this
area using computer based methods is found very little compared to the
magnitude of learning disability affected children.

The present method available to determine LD in children is based on

check lists containing the symptoms and signs of LD. This traditional method is
time consuming, not accurate and obsolete also. Such LD identification
facilities are much less at schools or even in cities. Parents are either unaware or

2 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

may not willing to take their children to undergo such an evaluation. Even if,
teachers are advised, parents may hesitate to such evaluation process because of
the unawareness of the society about LD as they might think that the child may
be mentally retarded. If the LD determination facility is attached with schools
and the check ups are arranged as a routine process, LD can be identified at an
early stage.

Under these circumstances, it is felt to design a tool based on machine

learning techniques for prediction of learning disability in school-age children.
Hence, it is decided to carry out a research work in the topic in a view to
increase the diagnostic accuracy of learning disability prediction. Based on the
statistical machine learning tool developed, the presence and degree of learning
disability in any child can be determined accurately at an early stage.

1.2 Problem Statement

The main problem considered in this work for analyzing and solving is
the design of a tool based machine learning technique for prediction of learning
disability in school-age children. The problem also involves in identifying the
important parameters of LD, identifying the hidden relationship between the
symptoms of LD and estimating the relative significance of each symptoms of
LD using data mining techniques. The drawbacks in the existing classification
algorithms have also to be determined first. Then how these algorithms can be
effectively modified and used in the prediction and classifications of learning
disabilities are studied. Thus the main task of the study is to find out how
effectively the different classification algorithms existing in data mining can
perform the prediction of learning disabilities. The problem also involves how
these drawbacks in the existing algorithms has determined while performing the
prediction of LD and how these are eliminated to maximize the accuracy of the

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 3
Chapter-1

optimized solution.

This research work has faced difficulties in collecting data from clinics
and schools. Either the doctors or the school authorities are not like to reveal
the data related to LD affected children. As there are no such related works in
the field, we felt difficulties in acquiring data as well as normalizing the
attribute values. However, we convinced the professionals in the field regarding
the necessity and effectiveness to be achieved, they agreed to sit along with
them during clinical consultations. Then from the experience gained with them
for the long term, we could evaluate the deficiencies in their traditional methods
of LD evaluation and thereby formed new informal checklists, which have been
approved by them and they are now using it.

1.3 Data Mining

In recent years the sizes of databases has increased rapidly. The amount
of data in the world seems to go on increasing and there is no end. In data
mining, the data search is automated by computer and stored electronically.
This has lead to the development of tools capable in the automatic extraction of
knowledge from data. The term Data Mining or Knowledge Discovery in
Databases, has been adopted for a field of research dealing with the automatic
discovery of implicit information or knowledge within databases [3]. Data
mining is defined as the process of discovering patterns in data. It is the non
trivial extraction of implicit previously unknown and potentially useful
information about data. It is an analytic process designed to explore data in
search of consistent patterns and/or systematic relationships between variables,
and then to validate the findings by applying the detected patterns to new
subsets of data [4]. The data mining process must be automatic or
semiautomatic. The patterns discovered must be meaningful in that they lead to

4 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

some advantage, usually an economic advantage [5]. Conventionally the

construction of model of the semantic structure of the dataset is the information
that is mined. The model might be utilized for prediction and categorization of
new data. Diverse fields such as marketing, customer relationship management,
engineering, medicine, crime analysis, expert prediction, web mining and
mobile computing besides others utilize data mining [6]. A majority of areas
related to medical services such as prediction and classification of effectiveness
of surgical procedures, medical tests, medication and discovery of relationship
among clinical and diagnosis data also make use of data mining methodologies
[7]. The kind of learning techniques that do not use the conceptual problems, it
is called machine learning.

The mining software is one of the analytical tools for analyzing data. It
allows users to analyze data from many different dimensions or angles,
categorize it and summarize the relationships identified. Technically, it is the
process of finding correlations or patterns among a lot of fields in large
relational databases. The patterns, associations, or relationships among all this
data can provide information. Data mining provide a link between the
transaction and analytical systems. Any one of the following four types of
relationships is generally mined for determining the existing patterns in the data
base. These relationships are classes, clusters, associations and sequential
patterns. Stored data is used to locate data in predetermined groups known as
Classes. Data items are grouped according to logical relationships are clusters.
Data can be mined to identify associations and sequential patterns are the data
mined to anticipate behavior patterns and trends. The three stages consists in
the process of data mining are initial exploration, model building and
predictions. Different levels of analysis available are artificial neural networks,
genetic algorithms, decision trees, rule induction and data visualization.

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 5
Chapter-1

Data mining has an inherent connection with statistics. Statistics studies

the collection, analysis, interpretation or explanation and presentation of data. A
statistical model is a set of mathematical functions that describe the behavior of
the objects in a target class in terms of random variables and their associated
probability distributions. Statistical models are widely used to model data and
data classes. Applying statistical methods in data mining is far from trivial [8].
When applying the high complex algorithm, data mining is required to
continuously handle fast, real time data stream. In general the statistical
methods are categorized in to two, viz. parametric and nonparametric methods.

Classification and prediction are the ultimate goals of data mining.

Classification is the process of finding a model that describes and distinguishes
data classes or concepts, for the purpose of being able to use the model to
predict the class of objects whose class label is unknown. The derived model is
based on the analysis of a set of training data.

In this research work, different classifications methods and models such

as neural network, decision tree, support vector machine, rough set, fuzzy set
and neuro fuzzy are used. The data sets are undergone to classification through
these methods or models. Then these models are used for prediction of learning
disabilities.

1.4 Statistical Machine Learning

Learning is an essential human property. It is the process by which
observed data is used for constructing models with the intention to use them for
prediction. Machine learning, a branch of artificial intelligence, is a scientific
discipline concerned with the design and development of algorithms that allow
computers to evolve behaviors based on empirical data, such as from sensor
data or databases. The general idea behind the statistical method is the detection

6 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

of class to learn a generative model fitting the given datasets and then to
identify the objects in low probability regions of the model as the class.
However in different ways we can learn generative models.

A major area of the machine learning research is to learn automatically

in recognizing complex patterns and make intelligent decisions based on data.
The difficulty lies in the fact that, the set of all possible behaviors, given all
possible inputs is too large, to be covered by the set of observed examples.
Hence the learner must generalize from the given datasets, so as to be able to
produce a useful output in new cases. Machine learning is concerned with the
development of algorithms and technique that allow computers to learn.
Machine learning has wide spectrum of applications including natural language
processing, search engines, medical diagnosis, detecting credit card fraud, stock
market analysis, etc. [6]. Some fundamental types of learning are supervised
learning, unsupervised learning, reinforced learning, etc. In the research work
different supervised algorithms and unsupervised algorithms are used for
predicting the learning disability accurately.

1.5 Learning Disability

The term ‘Learning Disabilities (LD)’ is a relatively new one. It was first
used by Dr. Samuel Kirk of Chicago, USA in 1963. The Children with Specific
Learning Disabilities Act (USA) was passed in 1969. However, it was not until
1990s that the biological basis for LDs found support. Specific learning
disability which includes dyslexia, dysgraphia, dyscalculia is commonly
referred to as ‘Learning Disability’ or ‘LD’ in India. Now special educators for
remediation are available in India [9].

A learning disability is found across all ages and socio-economic

classes. It is not a type of mental retardation as sometimes mistakenly thought,

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 7
Chapter-1

in fact, IQ scores could fall in the very high range [10]. LD is a hidden handicap
that affects academic achievement, vocational career and social life [11]. Every
child born in this world gets the care and comfort from the family in which he
or she is born. But the children with disability cannot enjoy and get such care
and comfort from the family [11]. If a child grows, he develops into a worthy
citizen. These children are able to quantify well and prove their worth.
Similarly the society expects the same from the disabled students also, here the
children failed. Specific learning disabilities have been recognized in some
countries for much of the 20th century, in other countries only in the latter half
of the century, and yet not at all in other places [12]. These may be detected
only after a child begins school and faces difficulties in acquiring basic
academic skills.

Learning disability is a general term that describes specific kinds of

learning problems. It is a neurological disorder that affects a child's brain and
impairs his ability to carry out one or many specific tasks. The LD affected
children are neither slow nor mentally retarded [12]. They have either normal or
above average intelligence. A child with a learning disability is often wrongly
labeled as being smart but lazy. A learning disability can cause a child to have
trouble learning and using certain skills. The skills most often affected are:
reading, writing, listening, speaking, reasoning and doing math. There is no
cure for learning disabilities [13]. There are also certain clues, most relate to
elementary school tasks, because learning disabilities tend to be identified in
elementary school, which may mean a child has a learning disability. A child
probably won't show all of these signs, or even most of them. They are life-
long. However, children with LD can be high achievers. They can be taught
ways to get around the learning disability. With the right help, children with LD
can and do learn successfully.

8 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

Learning disability is a disorder in which a child has difficulty in

learning in a typical manner, usually caused by an unknown factor or factors.
The unknown factor is the disorder that affects the brain's ability to receive and
process information [14]. This disorder can make it problematic for a child to
learn as quickly or in the same way as some child who isn't affected by a
learning disability. Learning disability is not indicative of intelligence level.
Rather, children with a learning disability have trouble performing specific
types of skills or completing tasks if left to figure things out by themselves or if
taught in conventional ways. A learning disability cannot be cured or fixed.
There are also certain clues, most relate to elementary school tasks, because
learning disabilities tend to be identified in elementary school, which may mean
a child has a learning disability. A child probably won't show all of these signs,
or even most of them. Even where they have been recognized, the amount of
help available varies from no services to their universal provision. This
unevenness in intervention services is tragic since most children with learning
disabilities who receive sufficient, knowledgeable remediation can proceed
through the school system and attain jobs that range from professor to labourer.
Conversely, if they are not helped, the possibility of adjustment of problems
arising is considerable. As our world becomes more complex, the knowledge
base increases and the concepts more abstract, an increasing number of children
will experience difficulty and be assumed to have learning to our collective
lives is not forfeited.

Learning disabilities are formally defined in many ways in many countries.

However, they usually contain three essential elements: a discrepancy clause,
an exclusion clause and an etiologic clause. The discrepancy clause states there
is a significant disparity between aspects of specific functioning and general
ability; the exclusion clause states the disparity is not primarily due to

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 9
Chapter-1

intellectual, physical, emotional, or environmental problems; and the etiologic

clause speaks to causation involving genetic, biochemical, or neurological
factors [15].

As many as 1 out of every 10 children in the United States has a learning

disability. Almost 3 million children (ages 6 through 21) have some form of a
learning disability and receive special education in school [12]. In fact, over
half of all children who receive special education have a learning disability
[15]. A learning disability often displays a cluster of characteristics over time,
in various intensities, which interfere with his/her overall development and
achievement. LD affected children can face unique challenges that are often
spreading throughout their lifespan. Depending on the type and severity of the
disability, interventions may be used to help the individual learn strategies that
will foster future success. Some interventions can be quite simplistic, while
others are intricate and complex. Teachers as well as parents will be a part of
the interventions. They can give aid to the children successfully in completing
different tasks. School psychologists quite often help to design the intervention
and coordinate the execution of the intervention with teachers and parents. With
the right support and intervention, LD affected children can succeed in school
and go on to be successful later in life. Social support is also a crucial
component for these type children in the school system and should not be
overlooked in the intervention plan. Parents of LD affected children often find
themselves attempting to cope with a bewildering array of problems. Their
children appear to be intelligent but they encounter all kinds of obstacles in
school [16].

In India the term disability is used synonymously as impairment, and

handicap or disability. These terms are different. The impairment means, the
loss of physical or sense organs. The child has not able to see, it is disability.

10 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

Handicap is the result of impairment and disability. Learning disability is a

broad term that covers a wide range of problems, including dyslexia and
behavioral problems and the full range of ability. If a child having learning
disability, that child requires special education needs. Learning disabilities vary
from child to child. One child with LD may not have the same kind of learning
problems as another child with LD [13].

This lifelong disability can interfere with the students’ acquisition of

academic and other basic skills necessary for survival as an independent adult.
Some of the common signs of learning disabilities and learning disorders in
children will be able to catch the problem early and take steps to get help to
child. It is very important in paying attention to normal developmental
milestones for toddlers and preschoolers. As early detection of developmental
differences is an early signal of a learning disability and thus the problems that
are spotted early can be easier to correct.

LD is real and a stumbling block for a nation’s development process.

The problems of children with specific learning disabilities have been a cause
of concern to parents and teachers for some time. When a LD is suspected
based on parent and/or teacher observations, a formal evaluation of the child is
necessary. A parent can request this evaluation, or the school might advise it.
Parental consent is needed before a child can be tested. Many types of
assessment tests are available. Child's age and the type of problem determines
the tests that child needs. A complete evaluation often begins with a physical
examination and testing to rule out any visual or hearing impairment [15].
Pediatricians are often called on to diagnose specific learning disabilities in
school- age children. Many other professionals can be involved in the testing
process.

The purpose of any evaluation for LDs is to determine child's strengths

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 11
Chapter-1

and weaknesses and to understand how he or she best learns and where they
have difficulty. The information gained from an evaluation is crucial for finding
out how the parents and the school authorities can provide the best possible
learning environment for the child.

Depending on the type and severity of the disability, interventions may

be used to help the individual learn strategies that will foster future success.
Some interventions can be quite simplistic, while others are intricate and
complex. Teachers and parents will be a part of the intervention in terms of how
they aid the individual in successfully completing different tasks. School
psychologists quite often help to design the intervention and coordinate the
execution of the intervention with teachers and parents. Social support can be a
crucial component for students with learning disabilities in the school system
and should not be overlooked in the intervention plan. With the right support
and intervention, children with learning disabilities can become great success in
school as well as later in the society.

1.5.1 Concept of Learning Disabilities

Learning Disability is a general term that refers to a heterogeneous group

of disorders manifested by significant difficulties in the acquisition and use of
listening, speaking, reading, writing, reasoning, or mathematical abilities. These
disorders are intrinsic to the individual presumed to be due to central nervous
dysfunction, and may occur across the life span [10] irrespective of regionwise.

Based on the frequency of occurrence, the important characteristics of

learning disability are

(i) Disorders of attention

(ii) Perceptual impairments
(iii) General coordination deficits

12 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

(iv) Disorders of memory and thinking

(v) Specific difficulties in the areas of reading, arithmetic, writing and
spelling and
(vi) Disorders of speech and hearing
Attention is the ability to concentrate on a task long enough to grasp its
essential features [11]. Learning disabled children are having short attention
span and lacking in concentration, but this does not mean that all inattenentive
children are learning disabled. A child may have an attention disorder for many
reasons. A significant characteristic frequently associated with learning
disability in a child is hyperactivity, although the relationship between the two
is still not clearly substantiated. Some expert refers to hyperactivity as an
attention deflects hyperactivity disorder. Hyperactivity is a much
misunderstood term and widely misused.

Children with learning disability are generally characterized by

inefficient memory systems. There is little doubt that in the area of academic
learning, children with learning disability are low achievers. They may have
problem in the specific areas of reading, writing and mathematics, although it is
not necessary that a child poor in spelling will also be poor in mathematics.
Listening is a complex process that requires good attention, discrimination and
memory systems. The child may use a lot of fillers such as articulation
difficulties. Learning disabled individual could make grammatical errors while
talking and use words incorrectly. Research in the area of learning disability has
primarily focused on assessment and diagnosis.

There is no one sign that shows a child has a learning disability. Experts
look for a noticeable difference between how well a child does in school and how
well the child could do, given his or her intelligence or ability. There are certain
general symptoms that may mean a child has a learning disability. A child

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 13
Chapter-1

probably won't show all of these signs, or even most of them. However, if a child
shows a number of these problems, then parents and the teacher should consider
the possibility that the child has a learning disability. Parents need to be
especially alert to developmental delays as a child approaches school age [16].

When a child has a learning disability, he or she may show some

symptoms. Some of these common symptoms are listed below [17];

(i) may have trouble learning the alphabet, rhyming words or connecting
letters to their sounds;
(ii) may make many mistakes when reading aloud and repeat and pause
often;
(iii) may not understand what he or she reads;
(iv) may have real trouble with spelling;
(v) may have very messy handwriting or hold a pencil awkwardly;
(vi) may struggle to express ideas in writing;
(vii) may learn language late and have a limited vocabulary;
(viii) may have trouble remembering the sounds that letters make or hearing
slight differences between words;
(ix) may have trouble understanding jokes, comic strips and sarcasm;
(x) may have trouble following directions;
(xi) may mispronounce words or use a wrong word that sounds similar;
(xii) may have trouble organizing what he or she wants to say or not be
able to think of the word he or she needs for writing or conversation;
(xiii) may not follow the social rules of conversation, such as taking turns,
and may stand too close to the listener;
(xiv) may confuse math symbols and misread numbers;
(xv) may not be able to retell a story in order (what happened first, second,
third); or
(xvi) may not know where to begin a task or how to go on from there.

14 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

1.5.2 Basic Types of Learning Disabilities

As pointed out earlier, learning disability is a lifelong disorder that

affects the manner in which, individuals with average or above average,
intelligence select, retain and express information. It reflects a difficulty in
encoding and decoding information as it travels between the senses and the
brain. Learning disabilities are also termed as ‘learning differences’, based on
the fact that certain individuals learn differently - they are not unable to learn,
but respond best to ways of learning that are different from traditional teaching
methods.

Many gifted and talented children are often misdiagnosed as having

learning disabilities or behavior disorders. This occurs because there are many
characteristics of gifted children, both social and emotional, that are mistaken as
symptoms of specific learning disorders. It is not uncommon for some gifted
children, with IQ scores over 140 to display a significant discrepancy between
Verbal IQ and Performance IQ and possess characteristics of a learning
disability. Often gifted children have unusual learning styles, and even though
they are very intelligent, they may also have learning disorders.

Learning disabilities tend to be diagnosed only when a child reach

school age. This is because school focuses on the very things that may be
difficult for the child — reading, writing and math, listening, speaking and
reasoning. Teachers and parents notice that the child is not learning as expected.
The school may ask to evaluate the child to see what is causing the problem.
Parents can also ask for their child to be evaluated. With hard work and the
proper help, children with LD can learn more easily and successfully. It's
important to remember that a child may need help at home as well as in school.

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 15
Chapter-1

A developmental lag might not be considered as a symptom of a LD

until the child is older, but we can intervene early if we recognize it when the
child is young. The activity of diagnosing the type of learning disability can be
overwhelming and time consuming. Try not to get caught up in trying to
determine the label or type of disorder and focus instead on figuring out how
best to support the child. The checklist, organized by skill set and age group,
can help in evaluating the child’s signs and symptoms and indicate whether
seek further assistance from a teacher or professional skilled in diagnosing
learning disabilities. Types of learning disabilities are often grouped by school
area skill set or cognitive weakness. If the child is in school, it will probably be
apparent if he or she is struggling with reading, writing, or math, and narrowing
down the type will be easier.

The common types of Learning Disabilities are explained below;

i. Dyslexia- Difficulty processing language- Problems in reading, writing,

speaking.

It is the most common type of LD affecting in children. It refers to a

specific difficulty in the area of reading. The terms generally used instead of
dyslexia are severe reading disability, primary reading disability, specific
reading disabilities and word blindness. It has been estimated that of the
children who attend school, approximately 10% to 15% have some
difficulty in reading and 85% to 90% percentage of all learning disabled
children have reading problems. Boys with reading problem outnumber
girls at the surprising rate of 4 to 1. In fact dyslexia has becomes
synonymous with learning disability to such an extent that it has been
suggested that dyslexia should be used as an umbrella term for all learning
disability in general. All learning disabilities are not dyslexia and the same
time dyslexia does not exemplify all learning disabilities. Dyslexia is a

16 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

broad term describing a reading disability. The word dyslexia implies the
meaning difficulty with words. It affects a child’s ability to read. It is
neurological in origin, likely to be present at birth and its effects are life-
long. It is characterized by difficulties with accurate and /or fluent word
recognition and by poor spelling and decoding abilities. These difficulties
typically result from a deficit in the phonological component of language
that is often unexpected in relation to other cognitive abilities and the
provision of effective classroom instruction [10].

Learning disabilities in writing can involve the physical act of

writing or the mental activity of comprehending and synthesizing
information. Expressive writing disability indicates a struggle to organize
thoughts on paper. Symptoms of a written language learning disability
revolve around the act of writing and include. They include problems with
neatness and consistency of writing, accurately copying letters and words,
spelling consistency, writing organization and coherence. Signs of a
language based learning disorder involve problems with verbal language
skills, such as the ability to retell a story and the fluency of speech, as well
as the ability to understand the meaning of words, parts of speech and
directions.

ii. Dyscalculia - Difficulty with math- Problems doing math problems,

understanding time, using money.

In 1919, Henschen reported that number blindness could occur

independently of specific reading disability [10]. Some persons observed
that memory and order disorders frequently occurred along with numerical
problems. Dyscalculia may result from lesions in widely disparate regions
of the brain. Dysfunctions associated within left temporal lobe were
characterized by difficulties with complex operations involving a sequence

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 17
Chapter-1

of steps or mental or oral calculation or reasoning. These disorders are

termed as secondary arithmetic disturbance. Children with learning
disabilities exhibit a variety of deficits in the area of mathematics. Shape
discrimination, size discrimination, sets and numbers and counting are the
different areas of dyscalculia.

Learning disabilities in math vary greatly depending on the child’s

other strengths and weaknesses. A child with a math based learning disorder
may struggle with memorization and organization of numbers, operation
signs, counting principles or have difficulty telling time.

iii. Dysgraphia - Difficulty with writing- Problems with handwriting, spelling,

organizing ideas.

Disorders of written language are referred to as dysgraphia and this

include difficulties in three areas of handwriting, spelling, content [10].
Many learning disabilities authorities have expounded why writing is so
important and how it can be taught to learning disabled students [18]. The
key characteristics associated with dysgraphia includes the main areas of
written language –handwriting, spellings and the content. Usually these are
interlinked problems and it is expected that a child having difficulties in any
one of these areas may experience a spill over in the others too. Most
disabled children hate to write and avoid it wherever it possible. Lack of
motivation becomes a real obstacle and a teacher needs to be at her creative
best when encouraging learning disabled child to write.

iv. Dyspraxia - Difficulty with fine motor skill- Problems with hand–eye
coordination, balance, manual dexterity.

Motor difficulty means problems with movement and coordination

whether it is with fine motor skills like cutting, writing, etc. or gross motor

18 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

skills like running, jumping etc. A motor disability relates to the output of
information from the brain. In order to run, jump, write or cut something,
the brain must be able to communicate with the necessary limbs to complete
the action. Signs that the child might have a motor coordination disability
include problems with physical abilities that require hand–eye coordination,
like holding a pencil or buttoning a shirt.

v. Auditory Processing Disorder - Difficulty hearing differences between

sounds- Problems with reading, comprehension, language.

The ability to hear things are greatly impacts the ability to read, write
and spell, but an inability to distinguish slight differences in sound or
hearing sounds at the wrong speed make it hard to explore words and
understand the basic concepts of reading and writing. Problems in visual
perception include missing slight differences in shapes, skipping lines,
skipping words, reversing letters or numbers, misperceiving depth or
distance or having problems with eye–hand coordination. In addition to the
above, the other disorders that make learning difficulty are anxiety, stressful
events, emotion, depression and other conditions affecting concentration
make learning more of a challenge.

vi. Visual Processing Disorder - Difficulty interpreting visual information-

Problems with reading, math, maps, charts, symbols, pictures.

vii. Attention Deficit Hyperactivity Disorder (ADHD)

It is a neurobiological disorder that starts early in childhood and can

continue into adulthood. The disorder is characterized by a delay or
permanent inability to self-regulate behavior or to control behavioral
responses.

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 19
 Chapter-1

viii. Social And Emotional Difficulties

Sometimes kids have trouble expressing their feelings, calming

themselves down and reading nonverbal cues, which can lead to difficulty
in the classroom and with their peers. Social and emotional skills are an area
where the parent can have a huge impact. For all children, but especially
those with learning disabilities, social and emotional skills are the most
consistent indicators of success, outweighing everything else, including
academic factors. Certain personal qualities and social relationships are very
much needed by the learning disabled children to make a satisfactory
adjustment to post school life [18].

1.5.3 Assessment and Remediation

LD can be a lifetime condition. There may be several apparent

overlapping learning disabilities in some children while others may have a
single, isolated learning problem that has little impact on their lives. LD is
diagnosed by a qualified child psychologist in association with a pediatrician.
The process of diagnosing a learning disability can be confusing. It shall be
started with the child's school. It involves testing, history taking and
observation by a trained specialist. A series of tests may be required to be done
to identify the affected areas. Special education brings some solution to the
problems the children affected with LD. Finding a reputable referral is
important. Since the educational needs of such children are different they will
be given special academic sessions in an integrated set up. Other professionals
such as speech therapist for children with speech disorder and physiotherapists
to those with motor deficits may be required. In some case psychotherapist may
also be involved in the treatment process.

20 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

When a LD is suspected based on parent and/or teacher observations, a

formal evaluation of the child is necessary. A parent can request this evaluation,
or the school might advise it. Parental consent is needed before a child can be
tested. Periodic evaluation of such children done to check for improvement will
help so that they can be integrated into the mainstream society as soon as
possible.

1.5.3.1 Assessment of LD

Assessment is the systematic process of collecting information about a

child, his past and current levels of performance his strength and weakness, in
order to help make better education decisions [10]. Assessment needs to be
relevant to the teaching goals and interventions that the child will receive.
Assessment is directly linked with how one will go about helping the child. It is
linked with intervention methods. The information collected through the
assessment must be relevant and of practical help in the class room. Assessment
helps parents to better understand their child’s problems and adjust their
expectations on the basis of the assessment data. It is pertinent to note that, in
India the history of LD assessment is still in its infancy.

Many types of assessment tests are available. Child's age and the type of
problem determine tests that child needs. Before any formal testing, a
conference is usually arranged between the child's parents and representatives
from the special education department. A factor that prevents accurate diagnosis
of twice exceptional is the prevalent practice of comparing gifted children with
the norms for average children. In psychology, as well as in other therapeutic
fields, such as audiology, speech pathology, occupational therapy and
optometry, the diagnostic question that is usually asked is how this child’s
performance compares with the norm. If the child scores within the normal
range, no disabilities are detected.

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 21
Chapter-1

The purpose of any evaluation or assessment for LDs is to determine

child's strengths and weaknesses and to understand how he or she best learns
and where they have difficulty. A major factor that makes it difficult to assess a
learning disabled child is the confusing nature of the disability itself. The
absence of testing instrument relevant to Indian students is another major
drawback. Most tests are designed for native English speakers and have items
which lie outside the cultural experience of the average Indian student.
Assessment can be expensive too.

Just as there are many different types of LDs, there are a variety of tests
that may be done to pinpoint the problem. Examples of the types of testing and
evaluations include:

(i) Interviews and direct observation of the child.

(ii) Review of the child's educational and medical history.

(iii) Parent conferences (a parent's input is of great importance because the

parent have observed the child since birth and can provide important
details about his or her growth and development).

(iv) IQ testing and psychological evaluation usually performed by a

psychologist or psychiatrist.

(v) Developmental history or social assessment.

(vi) Speech and language evaluation usually performed by a speech

therapist.

(vii) Evaluation of fine motor skills, visual-motor integration and sensory

integration usually performed by an occupational therapist.

(viii) Evaluation of gross motor skills, muscle tone and balance usually
performed by a physical therapist.

22 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

Two approaches are generally used to describe testing or the assessment.

They are formal and informal testing. The former uses standardized testing
while the latter uses non-standardized testing [10]. The standardized test
demands a high degree of uniformity in administration and interpretation. They
allow comparison of students of the same age or grade and can be use
individually or in groups. Informal tests or assessment are non-standardized
procedure used by teachers and other professionals to collect information. The
advantage of informal test is that they are simple to construct, administer and
score. They targeted at a specific objective and are of greater value for making
instructional decisions. The most frequently used informal procedures are
observations, interviews, questionnaires and tests. Check lists and rating scales
are another way of collecting behavioral data.

Generally, there are about 40 characteristics of LD, any one or more is

found in LD affected children. These characteristics in the form of a check lists
are used in the traditional methods of assessment of LD. These assessment
questions are formed based on discussions made with professionals/doctors
engaged in LD assessment fields. These 40 characteristics of LD, expressed in
the form of a questionnaire are listed below [17];

1. Do you find words difficult to spell?

2. Are you unsure about the use of full stops capital letters etc.?

3. Do you feel unsure about how to tackle reading unknown words?

4. Do you get very frustrated with your own performance?

5. Do you find it difficult to understand what you have read the first time?

6. Are you unsure how to organize writing a letter, a report or an essay?

7. Do you limit your writing to words you know you can spell?

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 23
Chapter-1

8. Do you feel you have to read every word on the page?

9. Can you spell a word one day and forget it the next?

10. Do you miss deadlines because you didn’t start early enough?

11. Do you find it difficult to listen and write at the same time?

12. Do you find it difficult to take message on the telephone?

13. Do you have difficulty in telling the difference between sounds?

14. Do you read slowly?

15. Do you have to ‘see’ and ‘feel’ if a word looks right?

16. Do you mix up dates and appointments so that you miss them or are
double booked?

17. Do you get numbers mixed up? e.g. roll numbers, telephone numbers?

18. Do you panic when you get to unknown words?

19. Do you confuse the order of the months of the year?

20. Do you remember words one day and forget them the next?

21. Do you find it difficult to follow written instructions?

22. Do you have days when things go dramatically wrong and equally days
when things go really well?

23. Do you find organizing your thoughts difficult?

24. When writing do you think that you end up with really what you wanted
to say?

25. Is map reading difficult?

26. Do you find it difficult to remember what you have read?

27. Do you find you have too many ideas to maintain focus?

24 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

28. Do you say the wrong word in the wrong place or at the wrong time?

29. Do you get the order of the letters wrong when writing?

30. Do you think your spelling weak?

31. Do you find organizing your life difficult?

32. Do you have difficulty telling left from right?

33. Does the print blur or move around as you are reading?

34. Do you find it difficult to skim and scan for information?

35. Is it difficult to see the detailed steps needed to complete a task?

36. Do you forget things quickly?

37. Do you have problems saying long words?

38. Do you lack confidence in yourself and think others are better than you?

39. Do you find it confusing putting sounds together to pronounce words?

40. Do you dislike reading aloud?

In the present research work, the method of informal assessment is

adopted for designing the tool for predicting the learning disabilities in children.
Even though different types of checklists are generally available for assessing
LD characteristics, a check list containing the 16 most frequent and important
characteristics (signs & symptoms) of LD collected from the above 40 general
characteristics, after eliminating the unwanted and redundant ones, is prepared
suiting to the LD conditions generally prevailing in Kerala. This general check
list adopted in this research work is shown in Table 1.1 below. As this check
list is prepared based on the experience of professionals including outside India,
the same is not region wise [24]. This check list is used as first phase of LD
assessment. Based on which, new check list is developed incorporating

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 25
Chapter-1

additional symptoms, viz. sub attributes, with different scores for each, as
discussed in Chapter 3. The new check list developed given at Table 3.3 is used
in the study of determination of LD, which can be used irrespective of region.
The increase in number of characteristics, if any, will merely increase the
prediction time.

Table 1.1 General check list

Sl. No. Indicators of LD Yes No
1 Difficulty with Reading
2 Difficulty with Spelling
3 Difficulty with Handwriting
4 Difficulty with Written Expression
5 Difficulty with Basic Arithmetic skills
6 Difficulty with Higher Arithmetic skills
7 Difficulty with Attention
8 Easily Distracted
9 Difficulty with Memory
10 Lack of Motivation
11 Difficulty with Study Skills
12 Does Not like School
13 Difficulty in Learning a Language
14 Difficulty in Learning a Subject
15 Slow To Learn
16 Repeated a Grade

The general check list can be easily used by the parents and teachers. It
is used to investigate the presence of learning disability. It includes the general
indicators of learning disability and focuses on understanding of the learning
disability. The goal is to provide concise and accurate set of diagnostic

26 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

attributes which can be implemented in a user friendly and automated fashion.

If a child has unexpected problems learning to read, write, listen, speak, or do
math, then teachers and parents may want to investigate more. The same is true
if the child is struggling to do any one of these skills. The child may need to be
evaluated to see if he or she has a learning disability.

The expert trained to do psychological testing and result interpretation

are clinical psychologists, school psychologists, educational psychologists,
developmental psychologists, neuropsychologists, occupational therapists,
speech and language therapists. Several professionals coordinate services as a
team to obtain an accurate diagnosis, including input from the child's teachers.
Recommendations can then be made for special education services or speech
language therapy within the public school system. If the public school is not
working out, then a nonpublic school provide specializes in treating learning
disabilities might be a good alternative. A professional learning disorders
specialist might refer to the importance of integration to learning.

1.5.3.2 Remedial solutions

Although several products are available for the identification and

remediation of learning disabilities, most of these are either unable to sustain
the progress of a disabled child or not aligned to government standards [19].
Pre-assessment helps separate the regular performers from the children who
have special needs. Parents and teachers can and should take an active role in
the child’s education.

All children can be both exhilarating and exhausting, but it may seem
that the child with a learning disability is especially so. Parents and teachers
may experience some frustration trying to work with the child, and it can seem
like an uphill battle. After the parents learn what their specific learning

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 27
Chapter-1

disability is and how it is affecting their behavior, they will be able to start
addressing the challenges in school and at home. A child with a learning
disability cannot try harder, pay closer attention, or improve motivation on their
own; they need help to learn how to do those things [20].

As a parent, discovering that something may stand in the way of the

child’s success can be unsettling and difficult. Whether or not the child has a
learning disability, remember that the way the parents and teachers behave and
what they do has the most impact on the child’s chances of success. Everyone
faces obstacles and the most important thing the parents and teachers can show
the child, apart from the consistent love and support and is how to deal with
obstacles. A good attitude won’t solve the problem, but it can give the child
hope and confidence that things can improve. The first task as the parent of a
child with a learning disability is to recognize that there are many things the
parents can do to help the child.

Parents often teach children to compensate for weaknesses, and gifted

children learn compensation strategies more quickly than their less capable
peers. What they neglect to tell the children is that compensation can break
down under various conditions. It takes more energy to compensate and when
one is fatigued, ill, stressed, dieting too strenuously, or adjusting to a new
situation, there may not be sufficient energy to support the compensation
strategy. So the individual is likely to experience good days when the
compensations work well, and bad days when they fail. They need to
understand that their high intelligence is revealed on their good days and that
there will be bad days, when their compensations, like bad brakes, fail to
support them.

Special education brings some solution to the problems of children

affected with LD [10]. They have to be given special academic sessions in an

28 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

integrated set up as their educational needs are different. Periodic evaluations of

such individual children have to be done, to check for improvement, which will
help them to integrate them into the mainstream society as soon as possible.
Nowadays, meeting the challenges in schools, most of the school authorities are
compelled to appoint resource persons/counselors in LD to diagnose the same
and attain early detection and remedies to improve the school atmosphere and
results.

1.6 Objectives of the Present Study

The following are the objectives of the research work:

i. Identifying the important parameters of LD using data mining techniques,

ii. Identifying the hidden relationship between the symptoms of LD,

iii. Estimating the relative significance of each symptoms of LD,

iv. Developing a new method for improving the accuracy of classifiers,

v. Developing fuzzy and rough set models in LD prediction; and

vi. Design a tool based on machine learning techniques for prediction of LD.

1.7 Scope of Work

The scope of research work includes the following;

i. Collection and preparation of real world data set,

ii. Knowledge extraction from domain experts,

iii. Prediction of LD with the help of machine learning algorithms,

iv. Identifying the frequent symptoms of LD,

v. Development of new approaches for LD prediction,

vi. Identification of the problems related to the classification accuracy,

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 29
Chapter-1

vii. Development of new algorithms for overcoming the identified problems,

viii. Application of new algorithms developed on existing classifiers,

ix. Developing new models in LD prediction with fuzzy and rough set
approaches

x. Development of an integrated knowledge based tool for the prediction of

learning disability,

xi. Determination of the performance of the developed tool; and

xii. Applicability and use of the developed tool in LD prediction.

1.8 System Framework

The main aim of the proposed research work is to design a tool based on
machine learning techniques for accurate prediction of learning disability in
school-age children and to effectively measure the percentage of LD present in
the child, according to the knowledge obtained from the clinical information.
The basic idea of the proposed knowledge based system is to detect the LD at
an early stage and to increase the accuracy of the learning disability assessment
and reduce the time consumed for the same. For achieving these goals, different
statistical machine learning technique are used. Towards the achievement of the
full objectives, a learning disability tool is constructed. The proposed tool has
many advantages compared to the traditional methods of learning disability
determination using check lists. These advantages include; consumption of less
manpower and time, the accuracy and efficiency underlying and handling of
missing values and redundant data. For the assessment of LD, check lists
containing the signs and symptoms of LD (attributes) are used. Obtaining a full
fledged information table by interviewing a child will not be a success in all
time as the same depends on the mood of the child at the moment of interview.
In such occasions, the data may contain missing values. To overcome such a

30 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

problem, missing value imputing by applying closest fit algorithm and

correlation based new methods developed are adopted in this research. Also
there are cases of most of the attributes in the data are unwanted and/or
redundant. To overcome this problem, dimensionality reduction method using
Principal Component Analysis (PCA) is applied.

The learning disability prediction tool is designed using the five

classification methods viz. modified neural network, modified decision tree,
fuzzy, fuzzy with reduced attributes and ANFIS. After applying the data
preprocessing on the data set, using closest fit algorithm or correlation based
method, as the case is, and PCA, the classification methods viz. decision tree,
fuzzy and ANFIS are applied.

In the developed tool, the knowledge obtained from the training is used
to predict the new data along with the presence of LD and its percentage. In
order to make the data appropriate for the mining process, it needs to be
preprocessed. In data preprocessing, the redundant data is removed, the number
of attributes is reduced and the missing values are imputed. The designed tool is
very user friendly. It provides student information record which stores the
details of the children in the student database, which can be retrieved as and
when required.

1.9 Thesis Contributions

This thesis makes several major contributions. Based on the statistical
machine learning tool developed, the presence of learning disability in any child
with its percentage can be determined. The class of LD like low, minor and
major and the percentage of LD in each class can also determined by this tool.
The research work provides the new insights into the interrelationships between
symptoms of LD, their relative importance and estimating the significance of

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 31
Chapter-1

each symptoms of LD. It contributes in developing new models of LD

prediction using fuzzy and rough sets and it succeeds in modifying the data
preprocessing with J48 decision tree and neural network for LD prediction. The
developed tool classifies LD as well as imputes the missing values in the data
set accurately. The work developed new algorithm based on correlation for
imputing missing values. The missing value imputing, done by the developed
tool contributes strong classification results. The number of attributes is
reduced by eliminating the unwanted ones by using PCA, helps in reducing the
time. The tool developed gives accurate results in lesser time compared to the
traditional assessment methods using check lists. The developed tool is very
effective for finding the LD affected child from the large database. This
research work has also considered an approach to handle learning disability
database to predict frequent symptoms of the learning disabilities in school
aged children. Early identification of LD will help the parents and school
authorities to recommend the child for early remediation, which will ultimately
help them to provide the child with best environment for his success. The study
will certainly contribute in the development of the nation as LD is a real
stumbling block for a nation’s development process.

Parts of this thesis work have been previously published as peer

reviewed journal papers and conference papers as shown in the list of
publications below.

1.9.1 List of publications

1. Julie M. David, Kannan Balakrishnan: A New Decision Tree Algorithm

for Prediction of Learning Disabilities, Journal of Engg. Science and
Tech., School of Engg., Taylor's University, Malaysia. (Accepted for
publication, 8(3), June 2013).

32 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

2. Julie M. David, Kannan Balakrishnan: Attribute Reduction and Missing

Value Imputing with ANN: Prediction of Learning Disabilities, Int. J. of
Neural Computing and Applications, Springer-Verlag London Limited,
DOI: 10.1007/s00521-011-0619, 21(7), Oct.2012, Online First, May
2011. Available at - http:// www.springerlink.com/ content/ evj556771
k561681/

3. Julie M. David, Kannan Balakrishnan: Prediction of Learning

Disabilities in School-Age Children using SVM and Decision Tree, Int.
J. of Computer Science and Information Technology, ISSN 0975-9646,
2(2), Mar-Apr. 2011, pp829-835. Available at - http:// www. ijcsit.com/
ijcsit-v2issue2.php

4. Julie M. David, Kannan Balakrishnan: Prediction of Key Symptoms of

Learning Disabilities in School-Age Children using Rough Sets, Int. J.
of Computer and Electrical Engg. 3(1), Feb. 2011, pp163-169. Available
at-http://www.ijcee.org/abstract/308-JE 351. htm

5. Julie M. David, Kannan Balakrishnan: Machine Learning Approach for

Prediction of Learning Disabilities in School Age Children, Int. J. of
Computer Applications, ISSN-0975-8887, 9(10), Nov. 2010, pp 7-14.
Available at - http://www.ijcaonline.org/archives/volume9/ number 11
/1432-1931

6. Julie M. David, Kannan Balakrishnan: Significance of Classification

Techniques in Prediction of Learning Disabilities in School Age
Children, Int. J. of Artificial Intelligence & Applications, 1(4),
DOI:10.5121/ ijaia.2010.1409, Oct.2010, pp111-120. Available at -
http://airccse.org/journal/ijaia/currentissue.html

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 33
Chapter-1

7. Julie M. David, Kannan Balakrishnan: Performance of Classifiers -

Significance of New Method in Missing Value Imputation and
Dimensionality Reduction: Prediction of Learning Disabilities in
School-Age Children, Int. J. of Applied Soft Computing, Elsevier B.V.
(communicated).

8. Julie M. David, Kannan Balakrishnan: Learning Disability Prediction

Tool using ANN and ANFIS, Int. J. of Transaction on Knowledge and
Data Engineering, IEEE. (communicated).

9. Julie M. David, Kannan Balakrishnan: Fuzzy and Rough Set

Approaches for Learning Disability Prediction, Int. J. of Artificial
Intelligence, Elsevier B.V. (communicated).

10. Julie M. David, Kannan Balakrishnan: Modified Pre-Processing

Methods with Machine Learning Algorithms for LD prediction, Int. J. of
Human Computer Studies, Elsevier B.V. (communicated).

11. Julie M. David, Kannan Balakrishnan: Prediction of Learning

Disabilities in School Age Children using Decision Tree, Proc: of the
Int. Conf. on Recent Trends in Network Communications, Lecture
Notes in Communication in Computer and Information Science, 90(3),
Springer- Verlag Berlin Heidelberg, DOI : 10.1007 / 978-3-642- 14493-
6_55, 2010, pp533-542. Available at-http://www.springerlink.com/content/
q 66788 rhk 77w7w15//

12. Julie M. David, Kannan Balakrishnan: Prediction of Frequent Signs of

Learning Disabilities in School Age Children using Association Rules,
Proc: of the Int. Conf. on Advanced Computing, ICAC 2009,
MacMillion Publishers Ltd., New York City, ISBN 10:0230-63915-1,
ISBN 13:978-0230-63915-7, 2009, pp202–207.

34 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

13. Julie M. David, Pramod, K.V.: Prediction of Learning Disabilities in

School Age children using Data Mining Techniques, Proc: of AICTE
Sponsored National Conf. on Recent Developments and Applications of
Probability Theory, Random Process and Random Variables in
Computer Science, Thrivikram,T., Nagabhushan, P., Samuel, M.S.,
(eds), 2008, pp139-146.

1.10 Road Map

The rest of this thesis is organized as follows;

A detailed literature survey is given in Chapter 2. Apart from the

introduction, the chapter contains detailed literature review on learning
disability as well as soft computing methods. The chapter ends with summary
and conclusion.

Chapter 3 is detailing about the data collection and implementation of

various soft computing methods. It begins with an introduction. The chapter
explains about technical background, data collection, data sets, data
distribution, knowledge extraction process, data normalization, attributes
normalization and entropy of LD attributes. The various implementation
methods adopted in this study such as neural network with back propagation
algorithm, decision tree with J48 algorithm, support vector machine with
sequential minimal optimization algorithm, bagging and fuzzy model are
explained well with comparison and results and pointed out about the insights
of LD. The chapter ends with summary and conclusion, and contributions.

Chapter 4 deals with improving performance by new pre-processing

methods. After the introduction, imputing missing values with closest fit
algorithm and with correlation based new algorithm developed, dimensionality
reduction with PCA, and modified data pre-processing and performance

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 35
Chapter-1

evaluation with MLP, decision tree, fuzzy model and neuro fuzzy model with
reduced attribute are dealt with in detail. The chapter also contains details of
study conducted on rough set model, apriori algorithm and clustering with k-
means algorithm. This chapter ends with comparisons and results, summary and
conclusions, and contributions.

Chapter 5 is development of an integrated knowledge based tool for LD

prediction. After the introduction part, the chapter contains system flowchart,
architecture of the tool and design of the tool. Tool testing, various screen shots
and performance evaluation of the designed tool are also covered in this
chapter. The chapter ends with summary and conclusions, and contributions.

The thesis concludes with Chapter 6, containing the summary of the

work and contributions of the overall study. This chapter also discusses the
future works and final conclusions.

1.11 Contemporary Works

There are only little studies available in the area of LD prediction with
knowledge based theories, as mentioned below.

Tung-Kuang Wu, et. al. in 2008, studied two well-known artificial

intelligence techniques, artificial neural network and support vector machine, to
the LD diagnosis problem [24]. This study is based on the formal assessment of
LD whereas the present research work relates to informal assessment of LD.
Maitrei Kohli and Prasad T.V., in 2010, proposed an approach for identification
of dyslexia and to classify potential cases accurately and easily by ANN [25].
As dyslexia is only a type of LD, the present research on general assessment of
LD is entirely different from their study.

36 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children
 Introduction

1.12 Summary and Conclusion

Learning disability is a classification including several disorders in
which a child has difficulty learning in a typical manner, usually caused by an
unknown factor or factors. The unknown factor is the disorder that affects the
brain's ability to receive and process information. However, children with LD
can be high achievers and can be taught ways to get around the learning
disability.

Individuals with learning disabilities can face unique challenges that are
often persistent throughout the lifespan. Depending on the type and severity of
the disability, interventions may be used to help the individual learn strategies
that will foster future success. Some interventions can be quite simplistic, while
others are intricate and complex. Teachers and parents will be a part of the
intervention in terms of how they aid the individual in successfully completing
different tasks. School psychologists quite often help to design the intervention
and coordinate the execution of the intervention with teachers and parents.
Social support can be a crucial component for students with learning disabilities
in the school system and should not be overlooked in the intervention plan.
With the right support and intervention, children with learning disabilities can
succeed in school and go on to be successful later in life.

When a LD is suspected based on parent/teacher observations, a formal

evaluation of the child is necessary. Parental consent is needed before a child
can be tested. Many types of assessment tests are available. Child's age and the
type of problem determines the tests that child needs. A complete evaluation
often begins with a physical examination and testing to rule out any visual or
hearing impairment. Many professionals can be involved in the testing process.
The purpose of any evaluation for LDs is to determine child's strengths and

Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children 37
Chapter-1

weaknesses and to understand how he or she best learns and where they have
difficulty. The information gained from an evaluation is crucial for finding out
how the parents and the school authorities can provide the best possible
learning environment for the child.

At present, the methods available for LD prediction and determination

are quite less and not accurate. The identification of such problems are really
challenging. In this research work, data mining using machine learning
techniques are used to analyze the symptoms of LD, establish interrelationships
between them and evaluate the relative importance of these symptoms. The
diagnostic accuracy of learning disability prediction is increased by way of a
knowledge based tool based on statistical machine learning. Identifying the
important parameters of LD prediction using the data mining techniques,
identifying the hidden relationship between the symptoms of LD and estimating
the relative significance of each symptoms of LD are also covered. The
developed tool has many advantages compared to the traditional methods of
using check lists in determination of learning disabilities.

JJJ

38 Statistical Machine Learning Techniques for the Prediction of Learning Disabilities in School-Age Children