Machine Learning

Quick & easy guide for every beginner

Sunny Kusawa
 Contents

Title Page
Preface
What is Machine Learning?
Why We need Machine Learning?
AI vs ML vs DL vs Data Science
Machine Learning Types?
Machine Learning Project Lifecycle
ApplicationS of Machine Learning?
How to learn Machine Learning
Stay connected with US
Afterword
Acknowledgement
 Preface
This book is for everyone who wants to get familiar with Machine
Learning.

It's articulated in such simple words,

graphics, and examples that anyone can easily understand it.

This book is for students, HR, managers, executives, non-technical

people, technical people, and everyone else, who are interested in
Machine Learning.

This book will help you understand the basics of Machine Learning,
what are different types of machine learning? how Machine Learning
project works and where exactly
Machine Learning is in use?

At the end the author also mentioned how you can learn Machine
Learning in the easiest way.

Let's start with this quick and easy guide for Machine Learning.
 About Author
Sunny Kusawa
On a mission to empower 10 million people in Artificial Intelligence,
Machine Learning and Data Science.

Passionate Data Science Mentor

Gold Medalist Software Engineer

9+ Years Software Industry Experience in Finance Domain

Youtuber - Talk about AI, ML and Data Science
https://www.youtube.com/c/DataMagic2020
Member of Board of Studies in Engineering Colleges

Writer on Medium.com

Passionate about learning, applying and sharing new

technologies

“Education is the most powerful weapon

which you can use to change the world."
Nelson Mandela

"A year spent in artificial intelligence is

enough to make one believe in God."
Alan Perlis

Index
1.What is Machine Learning
2.Why Machine Learning
3.AI vs ML vs DL vs Data Science
4.Type of Machine Learning
5.Machine Learning Project Life cycle
6.Applications of Machine Learning
7.How to build Career in Machine Learning
8.Connect with us
 What is Machine Learning?
Adding Learning ability in machine with the help of algorithms is Machine
learning.
Show data to machine and let machine learn different patterns from data with the
help of algorithms.

Here, data could be in any format like text, Records in table, images, videos, or
audios.
For example- Text data could be chat messages, email content, server log,
notification message etc.
Audio data could be songs, podcasts, recorded phone calls, recorded radio meets
etc.

Let’s try to understand with the help of an example. Take data related to any
specific task/object.

For example: take images of dogs and show them to machine. Machine will learn
different patterns about dogs with the help of Machine Learning algorithms.
The next time you show new dog image to a machine, it will be able to
understand its dog. Based on previous learning from dog images.

That's it, this is the way machine learn!

Why We need Machine Learning?
Machines can solve well defined tasks better than human beings.

AlphaGO- the Machine learning based computer program is the best

example.

In March 2016, Alpha go defeated 18 times world champion Korean Go

player Lee Sedon in Seoul
 Let's try to understand importance of Machine learning with another example.
Just imagine one doctor has 15 years of experience in cancer diagnostics.
 If we collect the knowledge of such 10 doctors and give it to a machine
to learn about cancer then obviously the machine will have more knowledge
than any one doctor.

This way with the help of data, machines are outperforming the human
beings. This is the reason machine learning in in use everywhere. It helps
human beings to solve complex problems better than humans.

Hope you got the taste of what Machine learning is and why we need it.
 AI vs ML vs DL vs Data Science
Most of the time beginner's get confused with what is
Artificial Intelligence, Machine Learning, Deep Learning, and data science?

Whether they are different from each other or the same?

If it’s different then how?

Let’s try to understand AI, ML, DL and Data Science one by one at a very high
level.

Artificial Intelligence is a broad field which includes Machine learning and deep
learning. On the other hand, we have a Data Science field which makes use of
ML and DL techniques for better data driven decision making.

Below picture represents the clear relationship between these fields.

 Artificial Intelligence
It is the field of Computer Science where we try to create/mimic humans
like Intelligence.
Machines can think and make decisions similar to humans with the help
of Artificial Intelligence.
Artificial Intelligence includes,

Robotics
Machine Learning
Computer Vision
NLP
Biometrics etc.

AI is a vast field, it has 4 types based on its capabilities,

Reactive
Limited memory
Theory of mind
Self-aware

It also categorized based on its intelligence level like,

Artificial Narrow Intelligence (ANI)

Artificial General Intelligence (AGI)
Artificial Super Intelligence (ASI)

Machine Learning

It is a sub field of AI. In this field, we try to add learning ability to machine where
machine try to learn different patterns and insights from data.
There are 3 types of Machine Learning

Supervised
Unsupervised
 Reinforcement

Learn more about machine learning types in detail with

examples in the upcoming section.

Deep Learning

It is a specialized Sub-field of Machine Learning which is inspired from

the human brain. It is more complex than Machine Learning and needs a lot
of computational power for processing. It deals with,

•Artificial Neural Networks

•CNN, RNN like Deep learning networks.

The evolution of deep learning is helping the Machine learning field to solve
complex problems with better performance but on the other side it’s also
responsible for the high carbon footprints.
 Data Science

Data science is all about playing with data for

getting insights and knowledge from data to solve business problems.
It involves different operations like,

Data Collection
Data Wrangling
Data Visualization
Data Analysis
Data Mining
 Data modeling and lot more.

Complex Data Science solutions use deep learning and machine learning
models for deriving the hidden patterns from the data.

Data Science allows businesses to make data driven decisions and grow
profits by offering better products/services to customers or by optimizing the
business process/resources.
 Machine Learning Types?

There are 3 types of machine learning based on how machines learn.

1. Supervised Machine Learning

2. Unsupervised Machine Learning
3. Reinforcement Machine Learning

Let’s understand each type of machine learning step by step.

1. Supervised Machine Learning

In supervised learning machines learn from the labeled data.

Labeled data means- when we pass data for learning along with that, we tag each
data with some label.

It helps machines know that this data is related to this tag/label.

For example: if we want a machines to learn about dog and cat then we pass
images of dog and cat to the machine along with a label assigned to each dog and
cat image.
It helps machines to learn about image and save its learning against that label/tag.

So, when a machine learns from an image and that image is labeled with CAT
then the machine saves that learning as CAT.

If we have 100 images of cats and those images labeled as CAT, then the machine
will learn the patterns from these 100 images and save these all learning for the
CAT label.

Supervised Machine Learning

Classification
Differentiating between different objects/records. Outcome is always discrete in
nature.

Ex. Classify given image is of dog or cat,

classify apple or orange,
classify cancer type from x-ray image

Regression
predict the continuous values for given input.

Ex. find weight of person based on age,

Predict the next years company revenue,
Predict next years employee churn rate

2. Unsupervised Machine Learning

In unsupervised learning machine learns from the

 unlabeled data.

Take data and pass it to a machine for learning and machine figures out the
different groups/clusters based on the similarity or common factors in different
data items

For example.
If you take your customer data and pass it to machine for learning without any
label

Then machine might create different clusters of

customers

based on the city they live in or

number of products they buy or
amount they spent buying products or
based on gender. etc.
 So here in unsupervised learning machine finds all common factors by
itself and creates clusters.

Unsupervised machine learning helps to derive patterns which are mostly

difficult to see by just looking at data.

For example: customer behaviour, market trends etc.

Unsupervised Machine Learning

Clustering
Group down the data based on similar features.

Ex. identify which are all products customers may buy together

Dimensionality reduction
Reduce the number of features in data by grouping down together.

For example:
you have 100 columns in your data. You can reduce the 100 columns to 40
columns by finding out which are all columns that have similar impact or no
impact on outcome.

It helps to save memory and computational power required for processing the
high dimension data.

3. Reinforcement Machine Learning

In reinforcement learning machine learns by rewarding desired behaviors and

punishing undesired one.

Let's try to understand an example of self-driving cars.

A car driving on the road by itself. It can possibly take one action - left Turn,
Right Turn or NO turn

Based on action, it will get reward or punishment.

The desired behavior is that the car should run fine without hitting other cars
and road borders.

Keep taking actions and keep getting rewards and punishments. It learns from
punishments and tries to Improvise on every action.
This way the car learns to drive by itself and this way of learning we call it
reinforcement learning.
 Machine Learning Project Lifecycle

Step by step process of Machine Learning Project

1.Understand Business Problem

2.Collect Data
3.Prepare Data
4.Train Model
5.Evaluate Model
6.Test Model
7.Deploy Model
 Lets try to understand each step in shorts,

1.Understand Business Problem

In this step understand what is the exact business problem?

Why company want to solve this business problem?

Who is getting impacted by this problem?

Can we solve this business problem with Machine Learning?

Where can we get the required data to solve this business problem?

If this problem can be solved with Machine learning and data is also available or
possible to acquire from some source. Then move to the next step.
 2.Collect Data

Data required for a Machine Learning project can come from different
sources depending upon what kind of problem you are solving.

Company own data

Acquire data from third party
Public/Open-Source data
Extract data from online sources- Web scraping
Or any other source
Some times if there is lack of data from one source then data is collected
from multiple sources to build one good size data set.
3.Data Preprocessing

Machine learning algorithms won't understand raw data most of the time
or raw data won't enable ML algorithms to perform at its best level.

In the data preprocessing stage convert the data in such a way that
machine learning algorithms can understand it better and be able to learn
best from it.

Below is the possible task performed in data

preprocessing
Handle outliers
Handle missing values
Handle categorical variable
Data scaling
Handle imbalance dataset
Data binning
Dimension reduction
 After Data Preprocessing Split data into 3 parts.

Train data
Validation data
Test data

For training\learning, validation and testing

purpose

4.Train Model

Train Multiple models with preprocessed data.

Training means allowing machines to learn from data.

Model means Program which is learning from data and saving knowledge.

See which model works best with training data.

Select the model which is working best. Refine/tune this model so it can give
better performance. It's called hyper parameter tuning.

5.Evaluate Model

Check models performance on different metrics/measures like accuracy,

time Taken, positive result ratio, negative result ratio etc.

In technical terms these measures are called as precision, recall, f1 score,

AUC curve etc.

In short, here we cross verify the performance of the model.

 6.Test Model

Test the trained and evaluated Model with test data

Test data is data which is not yet seen by model. So, after showing new
data (test data), check how model is working.

If the model works well on test data it means the model is a good one
and ready to serve.

7.Deploy Model

deploy this model on local machine, private server or on cloud as per business
requirement.
In this step Model is made available for business so their
clients/customers/employees can use it

Important:

During the process of Machine Learning project development. It's required to

present the results and data analysis to clients/customer/higher management as
and when required. Data storytelling is very important while interacting with
stakeholders.

Keep repeating:
After deployment keep monitoring machine learning models. If it starts
misbehaving or starts giving wrong results then keep repeating all steps
again starting from understanding the problem, collecting data till model
building and deployment.

It's a continues process which keeps repeating whenever ML models

start performing poorly.
 ApplicationS of Machine Learning?

Weather prediction
Machine learning used to forecast wheather.

Ecommerce sites like Amazon

Uses Machine learning for personalized landing page, special offers, delivery time estimation,
product recommendation etc.
Uber/ OLA Cabs
Uses Machine learning for best route selection, on demand pricing, estimated time to reach destination.

Robots
Uses Machine Learning for identifying objects, understanding human voice, taking actions based on
situations etc.
Disease Detection
Uses Machine learning for Cancer detection, Clinical trials, robotic surgeries etc

Entertainment
Uses Machine Learning for Netflix movies recommendation, YouTube ads, YouTube auto playlists etc
Agriculture
Uses Machine Learning for plant disease detection, plant growth tracking, fertilizer spreading through drone,
drone surveillance, weather prediction etc

These are just a few examples to give an idea about where Machine learning is getting used and
how.

Machine Learning is in,

Finance,
Social media,
Robotics,
Entertainment, Automotive,
Education,
E-commerce,
Agriculture,
Transport,
Healthcare,
Astronomy,
Gaming,
Data security etc.

Now you name any field and machine learning has its footprint over there.
 How to learn Machine Learning
If you are trying to learn Machine learning with self study then below is the
easiest way to learn in my opinion. This is a very high level flow which you can
follow,

Learn Machine Learning basic theory. What is Machine Learning, its type,
different kinds of ML problems, ML project life cycle, a couple of basic ML
algorithms etc.

Follow Machine Learning communities like kaggle, Reddit, Facebook Groups

related to AI, ML and Data Science etc.

Learn basics of python or R programming language which will help you in

implementation of Machine learning projects. I would recommend Python
because you will get alot of free resources and open source API's.

Download some most popular machine learning project code notebooks from
the Kaggle community. Try to run it on your system/google colab, It will give you
good confidence.

Try to do reverse engineering and see how different machine learning project
lifecycles steps applied in the project code.

Try to explore more Machine Learning projects and keep learning new
concepts, ML algorithms and techniques as you come across. if you aren't able to
understand any code snippet then copy and search on Google. You will get lots of
resources.

Take some sample business problems, collect data and try to apply the
techniques which you learnt from previous project notebooks.

Keep repeating this process and try to do different kinds of projects. Each
time you will end up learning a couple of new concepts and techniques.

Don't focus too much on maths and statistics at beginner level. It might be
overwhelming at the initial stage.
 Focus on applied Machine Learning.

Learning by self study sometimes becomes difficult. You may get stuck
somewhere or you may lose interest in between or you may won't able to apply
yuor learnings on new business problems etc. Looking at these issues we have
designed a Machine Learning in detail course with different business case
studies and hands on experience.

If you want to develop good Machine Learning skills in 3 months then you
can opt for our Applied Machine Learning course. You can contact on below
email address for more details.

Contact us:
datamagic2020@gmail.com

You can also click on below contact button and fill the Google form to
directly connect with us,

Contact
Stay connected with US
If you want to learn more about Machine Learning, Artificial
Intelligence and Data Science then join me on below platforms,

YouTube
Facebook Page
Facebook Group
Instagram

New Machine Learning book with business case

studies and in depth concepts with coding.
coming soon…
 Afterword
Hope you enjoyed and found this Quick and Easy Machine Learning
Guide helpful. If yes then please provide your review comments on
Amazon.
 Acknowledgement

I would like to thank my mentor Dev Gadhavi, Sugeeta Madan and Sushant
Kumar for motivating me to start writing a book.

A special thank to my family members who supported me in this journey.

Thank you universe for helping me in writing my first e-book.