Introduction to machine learning

Noel Moses Mwadende is currently 2 year Computer Science and
Information Security student at University of Dodoma , Data Scientist and
well specialized in python for ethical hacking ,Penetration Tester, Malware
Analiysist, Hacking tools writer already written “INFORMATION
GATHERING TOOL” mostly used for penetration testing phases of
Information Gathering and Reconnaisance, writer of different books about
programming language, hacking, malware and Computer security in
general also a member of Udom CyberSec lab, After making several
experiments in Udom CyberSec lab , Noel started to write different books
concern security with great passion starting with his book of "WIFI
HACKING IN FEWS STEPS" ,"PANDAS TOOL FOR DATA
SCIENTIST","HOW TO MAN IN THE MIDDLE ATTACK",”MAKING
OUR ANDROID TROJAN-HORSE”,”MY WIN TROJAN-HORSE” and
"WRITE VIRUS BY BATCH PROGRAMMING" and still different books are
being written , and I will keep on releasing my different written books for
you my reader , but don’t get tired.
Introduction To ML
Machine learning
Noel Moses Mwadende
TANZANIA MoTech

ABOUT THE AUTHOR
Noel Moses Mwadende is currently 2 year Computer Science
and Information Security student at University of Dodoma, Data
Scientist and well specialized inCyberSecurity.
Apart from this books other books written by same author ,
which are present at Github https://github.com/MoTechStore
are "WIFI HACKING IN FEWS STEPS" ,"PANDAS TOOL FOR
DATA SCIENTIST",”ZIP PASSWORD CRACKER” and
“INTRODUCTION TO MACHINE LEARNING”.
Currently the author has been employed at MoTech YouTube
channel (
https://www.youtube.com/channel/UCtuaigKZF3okQnKON5R
M1qQ/playlists ), and so far the author have produced more than
70 tutorials about python, machine learning, pandas, VB.net, php
and other Information Technology.

INTRODUCTION
This book covers basic concepts for beginners in machine
learning, explaining different types of machine learning,
algorithms consisted in each type of machine learning, and how
to do machine learning.
Inshort the intention of this edition just to make awareness and
derive this new concept, because it happens in most cases, some
beginners of machine, just start with learning algorithms, while
ignoring the basic concepts which are basic for their field.

ACKNOWELDGEMENT
I would like thanks to my instructors of machine learning
and Data Science Sir Salim Diwani, Mr Eliah and Miss Martha
Shaka, to be honest these are the most import people in my ,
because I can not am good data scientist without mention their
names, they made forgotten.
Thanks to my security instructor Sir L Mutembei, your role
in my development in writing books, producing tutorials at
MoTech, and of machine learning for cybersecurity.
Thanks to MoTech for sponsoring of the writing of this
book to be host without them for sure I could prove some
failure in writing this book.
Thanks to Kelvin N Simchimba currently MTA 2 year
student at UDOM, though I contacted you so late but you
made good design of cover page and the whole structure of the
book.
Lastly, I would like to thanks my friend JohnBosco Francis
because my first lecture to you about machine learning
introduction was the starting point of this book, because after that
session then I thought why can’t I teach others about machine
learning, how ?, it is through this book was made.

Table of Contents
CHAPTER ONE ....................................................................................................................................
1.0 INTRODUCTION OF MACHINE LEARNING …………………………………….6
1.2 COMMONLY MIXED TERMS IN MACHINE LEARNING ……………...............7
1.3 RELATION BETWEEN DATA SCINCE AND MACHINE LEARNING………11
CHAPTER TWO .........................................................................................................................................
2.0 TYPES OF MACHINE LEARNING …………………………………………………13
2.1 SUPERVISED MACHINE LEARNING……………………………………………..14
2.2 SEMI-SUPERVISED MACHINE LEARNING ……………………………………..16
2.3 REIFORCEMENT MACHINE LEARNING ………………………………………..18
CHAPTER THREE .............................................................................................................................
3.0 STAGES OF MACHINE LEARNING PROJECTS…………………………………………………………………23
CHAPTER FOUR........................................................................................................................
4.0 HOW MACHINE LEARNING WORKS................................................................. 26
CHAPTER FIVE..........................................................................................................................
5.0 APPLICATION OF MACHINE LEARNING .................................................... 28
CONCLUSION ...................................................................................................................................... 29
REFERENCES ........................................................................................................................................ 30

INTRODUCTION OF MACHINE LEARNING
Machine Learning is the system that can learn from
examples through self-improvement and without explicitly
coded by programmer, the meaning come with idea that
machine can learn from data or examples to produce accurate
results. Machine learning is closely to data mining and
Bayesian predictive modelling.
Simple example where machine learning can function as
human being, consider this scenario, for old man lived 85 year
, coming across through different summer and winter seasons
can predict the occurrence or thee rate of rainfall in the next
year, how this becomes possible ? it is due to experience, also
for machine learning given a Dataset about rainfall statics
organized in good way, also computer can learn from those
Dataset and come up with prediction.
A football fan haven watching Manchester United Vs
Liverpool games for 10 years, have data, experience, and can
predict the game who is going to win the match, this is same
for machine learning, by using different machine learning
algorithms and Dataset, the system can be made possible to
predict.

Arthur Lee Samuel (1959) described machine learning as "It is the
field of the study that gives the ability to the computer for self-
learn without being explicitly programmed ". machine learning,
learn itself from Dataset exposed to them and finally they come
with great output.
It good view from Arthur Lee Samuel, but if machine learning
project is deployed by yourself it will need programming
language concepts, since a lot of people nowadays are using PHP,
Django and Flask for deploying their projects.

COMMONLY MIXED TERMS IN MACHINE LEARNING
Data science
is a multi-disciplinary field that uses scientific methods,
processes, algorithms and systems to extract knowledge and
insights from structured and unstructured data. Data science is
very wide field as it covers a lot of things, for someone to be
specialist in data science will need a lot of combination
knowledge like programming language, idea about
mathematics and statistics, database, it also involve machine
learning.
Machine Learning
is the field of computer science that deals with giving ability to
the computer to self-learn and do a work which may be prediction
or any other intended work .It will need Dataset which will be
exposed to the algorithm, for it to learn, in supervised machine
learning Dataset are usually labelled, algorithms learning from
features containing names, but this is quite different to
unsupervised machine learning where Dataset have no labels, so the
algorithms learn independently.
Big Data
is the field that deals with the ways to analyse, systematically
extract information from data set that are too large or complex to

deal with by the traditional data processing application software.
Data with many rows offer greater statistical power, while data with
higher columns may lead to higher false discovery rate.
Data Mining
is the process of discovering patterns in large data sets
involving methods such as machine learning, statics and database
systems.it is an interdisciplinary subfield of computer science and
statics with an overall goal to extract information from data set and
transform the information into a comprehensible structure for
further use .Different algorithms are used to extract data from social
media, which in turn is used for health centres and production of
products according to the desire of customers.
Algorithms
are the methods or procedures used to get the task
done or to solve a certain problem sometimes these algorithms are
called classifiers.
These algorithms are already developed, the remaining task is to
import and use them. Sklearn is library which contains machine
learning algorithm.
Models
are well defined computations formed as the result of an
algorithm that takes some values as input and produces some value

as output .so model is the result of algorithm, they take some values
as input, process them and come with output.

RELATION BETWEEN DATA SCINCE AND MACHINE LEARNING.
Data science
is wide term, covering many fields like machine, statics, analysis and
some concepts of programming language. machine learning is direct
term which involve algorithms, to be data scientist is not necessary
to have knowledge of machine learning, but also for good data
scientist will need you to cover machine learning.
Machine learning
includes some techniques that can be useful for data scientist, but
data science does not more rely on machine learning, but data
science is much more broad or wide.

Introduction to machine learning

TYPES OF MACHINE LEARNING
1. Supervised
2. Unsupervised
3. Semi-supervised
4. Reinforcement

1. SUPERVISED MACHINE LEARNING
is the type of machine learning that learn from past input
data and make future prediction as output.is the machine learning in
which labelled data used to train the algorithms, algorithm are
trained using data with labelled names where the data is divided
into train and test same like input data and output data, these input
data are called features, indicated by x and output data which is
indicated by y.
SUPERVISED MACHINE LEARNIG METHODS
A. Classification (yes or no, High or low)
B. Regression

A. = = => Classification (yes or no, High or low)
This is concerned with building models that
separate data into distinct classes, it always deals with problem
which have binary output, that is two output.
Classification Algorithms
= Logistic Regression
= K-Nearest Neighbours
= Support Vector Machine
= Kernel Support Machine
= Naive Bayes
= Decision Tree
= Random Forest

B. = = => Regression
This is based on taking input data and then machine learning
predict continuous output values, as you can see that the data
differs between classification and regression learning, in
classification the output is binary but in regression learning the
output is continuous.
Classification Algorithms
= Simple Linear Regression
= Decision Tree
= Random Forest
= Support Vector Machine

SEMI-SUPERVISED MACHINE LEARNING
It uses small amount of labelled data and large amount of
unlabelled data this is contrary to supervised machine learning
which uses complete labelled data. It is mostly used for speech
recognition and classification of texts.
Semi-supervised Algorithms
= Graph Based Algorithm
= Generative Models
= Self Training

REIFORCEMENT MACHINE LEARNING
Is this machine learning, model are able to learn based on reward
and punishment receive for it’s previous actions, it make decision
sequentially according input and out to the model.
Reinforcement Machine learning Algorithms
= SARSA
= DDPG
= Deep Q Network
= Q- Learning

2. UNSUPERVISED MACHINE LEARNING
is the type of machine learning that try to find hidden
structure pattern by using unlabelled data, the model is given
data or examples then after understanding the new data is
given to the model to test it.
UNSUPERVISED MACHINE LEARNING METHODS
A. Clustering
B. Association

A. Clustering
This is used for analysing and grouping data which are not
labelled class or even a class attribute at all.it is algorithm which
deals with grouping sets which are similar into clusters.
Clustering Algorithms
= K-Means
= Hierarchical clustering
= Hidden Market model
= Fuzzy C-Means

B. = = => Association
Discover the probability of co-occurrence, how can multiple
item occur together, how the occurrence of one item have an
association to the occurrence of another. Most employed for
recommendation systems and arranging of products in
supermarket.
For example, pen, pencil and exercise book can be associated
together, mil and bread can be associated or grouped together.
Association Algorithms
= Aprior
= Eclat

STAGES OF MACHINE LEARNING PROJECTS
1. Data Collection.
2. Data Preparation.
3. Choose a model.
4. Train the model.
5. Evaluate the Model.
6. Parameter Turning.
7. Make Predictions.

1. Data Collection.
This step involving getting or finding data which will be used for
your project, one can get data direct from organization, but if it is
normal research consider your selected area of research ,for
beginner to machine learning there are many repository UCI
,Kaggle ,Google Dataset research and visual data containing a
huge amount of data.
2. Data Preparation.
This step will involve cleaning of data, and changing them to
numeric type, so if data contain question mark, quotes,
punctuation should be removed, removing of extra white space,
removing duplicated data, and making all missing values are
removed by either filling them or totally removing them.
3. Choose a model.

Criteria for choosing a model is according to the problem you
solve, for example if machine learning is concerned with output
which is binary , regression algorithms like logistic regression
will be used, but if machine learning is about predicting
continuous values like price linear regression will be used, but
sometimes the prediction may be multiple where one can choose
to use random forest.
4.Train the model.
Divide Dataset into train and test, almost train and test data are in
ration of 0.7:0.3, most of the time training data is 70% and testing
data is 30%, then after data is divided, fit the model.
5. Evaluate the Model.
Checking the result of model, cross validation is used evaluate
performance of the model, where performance of the model can
be cross validated according to the value of k (k fold).
8. Parameter Turning.

These are made internally in algorithm, that are required when
using a model, for example in linear regression and logistic
regression coefficient is their parameter and support vector in
support vector machine, so these parameters usually need to be
passed in algorithms for it to function well.
According to different algorithms there are parameters which
are internal configure in the model, but there is hyperparameter
which are external to the model, this part is about optimizing a
model.
9. Make Predictions.
After everything is set clear use it for making prediction.

HOW MACHINE LEARNING WORKS
It is like the way machine learn is like human being,
human learn from experience, as long as we know is when our
ability to handle different situation grow, it is where we can
predict different things in the society, for example an old man
who has lived about 125 years , passed through many summer
and winter season in his lifetime can easily predict when will
the rain start for the next year due to his experiences, it is the
same way machine learning works , given many data
composes of many rows and columns about rain statics of the
past year it will become easy for it to predict when or at what
amount the rain will be.

APPLICATION OF MACHINE LEARNING
1. Search engine result
2. Voice recognition
3. Face recognition
4. Prediction
5. Spam filtering
6. Recommendation system
7. Social media
8. Sentiment analysis
9. Video surveillance
10. Speech recognition

CONCLUSION
Thank you that was the end, In next edition will be about
supervised machine learning algorithms, this was just
introduction, it is better you understand all basic, so that when
comes an issue of choosing algorithm, you should know when
and where to use the model, and why.
Digging deep into this book will turn you into one of Best in
machine learning, for sure it will move you to a certain level, for
those who takes data science, python and machine learning
https://www.youtube.com/channel/UCtuaigKZF3okQnKON5R
M1qQ/playlists ), is the link you can get my self made tutorial
about python, pandas and machine learning, but also
https://github.com/MoTechStore is link to get my previous
books, and will keep on adding books , welcome and review.

REFERENCES
1. Duda, R.,Hart and Stock,D.(2001). The Elements Of
Statistical Learning – Data Mining, Inference and
Prediction.Berlin:Springer-Verlag.
2. Mitchell,T.(1997).Machine Learning. New York: Mc Graw-
Hill.
3. Bishop,C.M.(1995). Neural Networks For Pattern
Recognition. New York: Oxford University Press.
4. Baldin,P. and Brunak,S.(2002). Biometrics: A Machine
Learning Approach, MA: MIT Press.
5. Goodfellow,I.,Bengio,Y and Courville,A.(2016). Deep
Learning. MIT Press
6. Christianini, N and Shawe-Taylor,J.(2000). An Introduction
To Support Vector Machine. London: Cambridge University
Press.
7. Vapnik,V .(2013).The Nature Of Statistical Theory. Springer
Science & Business Media.
8. Jollifie, L. (2002).Pricipal Concept Analysis.Wiley Oline
Library.
9. Tan, P-N., Steinbach., and Kumar, V.(2004). Introduction Ta
Data Mining. New York Addison-Lesley.
10. J.O.Berger,.(1985). Statistical Decision theory and
Bayesian Analysis,Springer.New York
11. C.R.,Rao.(1973). Linear Statistics inference and it’s
application,John Wiley and Sons.New York
12. G.R. Shorack and J.A Wellner. (1986).Emeprical Process with
application to Statistics,Wigley.New York

Introduction to machine learning

More Related Content

Similar to Introduction to machine learning (20)

More from MoTechInc (7)

Recently uploaded (20)

Introduction to machine learning