Machine Learning from Zero to One

HOW TO BECOME DATA SCIENTIST OR MACHINE LEARNING ENGINEER SELF STARTER WAY WITH ZERO COST.

Hello all Rajendra here. In this blog we will talk about how to learn machine learning from zero to one without any prior knowledge in a self -starter way. Yes without zero experience in machine learning you will learn a Machine Learning .This blog is for those who want to start their journey towards Machine Learning or want to learn Machine Learning and become a super hero of world’s most game changing technology. These days everyone is talking about Machine Learning and today Internet is flooded with one query — “ How to learn Machine Learning?”

We are living in an era where technology changes impossible things to possible day by day. Today technologies like Big data, IOT, Cloud and AI along with Dev Ops culture, which emerged in last few year’s changed the world and made impossible things become a reality beyond the imagination. These technologies are booming at tremendous speed nowadays. AI has become a new buzzword today. Lot of people are struggling to get up skilled and gain knowledge on these technologies. There is a lot of material available on Internet today, but people are struggling to find out proper learning path in order to learn these technologies. Also people are spending lot of money without thinking in a practical manner or knowing about the ground reality.
One thing lot of people are not aware of these technologies and not sure how to use it in real world and how things work in real. Everyone today uses AI/ML directly or indirectly. Tech giant companies are developing cool products with these technologies and changing lives of millions in amazing way. But why are these technologies booming right now? Why is everyone talking about these technologies right now? Have they emerged recently, or existed in the past?
These technologies boomed in a recent years. The reason behind this is data which is generated today is huge and the velocity of its generation is much faster compared to the last 10 years. Also due to cloud technologies today we have lot of resources for processing this data well and find out the meaningful pattern from data and make some decision based on that pattern. Data is oxygen for AI. Lot of companies today processing this data which is either structured or unstructured and deriving meaningful patterns from it and running their business or developing great products for human
community. So Data is most important thing right now for all companies.

So after this short introduction or background of these technologies let’s start journey towards learning AI.

Welcome on-board…

A lot of people have inhibitions about Machine Learning like — there is a lot of maths needed for machine learning and hence it is difficult. This is a total myth. Machine Learning surely requires mathematical and statistical skills but definitely doesn’t require you to have an expertise in the same. Having a background about statistics and Maths will just makes things simpler for you.
Before starting with Machine Learning let’s first start with Data analyst and Data Science.
In order to start with ML, one must first invest time in understanding nature of your data. Meaning, Data is most important thing for AI. In order to build a Machine Learning application, first you need to prepare your dataset. In order to do that you need to do lot of stuff with data which we called Data Analysis (EDA).

How do we do Data Analysis?
Almost 80% time of the overall project life cycle of ML application goes into
preparing correct and meaning full data set based on which we train our model. In order to do data analysis as first step and for that a lot of tools like SAS, Tabula and R studio and many more are available in the market.
Programming languages like R, Python and many more can be used for data analysis. Most Data scientists however use Python as primary language for Data analysis and model building . People uses the R for same.

In order to learn data analysis using python one should first aware of 
python and below packages/libraries of python.

#1. Programming Foundation:
1. Numpy
2. Pandas
3. Matplotlib
4. Scipy
5. Sklearn
6. Seaborn

There are a lot of other packages one may use as per requirement, but this is good for beginners. These libraries help in data visualization and plotting of data which helps in understand the nature of data, feature selection and all.
Here is a list of some very useful links for learning Python. In order to start with Python from basics to advance check below You Tube links.
1. https://www.youtube.com/watch?v=apACNr7DC_
2. Check Google developer class on python.
3. Python for Everybody coursera.
4. Python Programming Foundation Udacity
5. Introduction to computer science using Python by Harvard

All courses are available on Internet.Below are the best books on Python:
1. Learn Python the Hard way
2. Learning Python
3. Python cookbook
4. Python for Data analysis
5. Python for Everybody
6. Automate the boring stuff with Python.

Once you get hands-on over Python next step is to create an account on
Datacamp and start practice of above python libraries .

Special thanks to Udacity, Coursera, Edx, Datacamp and Udemy.

#2. Data Analysis and Data Science Foundation:

Once you are aware of Python programming and have a hands on experience on the same, you may move towards data science and data analysis. I will recommend spending at least 2 hours minimum on practicing coding make you good coder in 2 months. It will help you develop your logical skills and transform ideas into code. Also try building small applications using Python.

Please find a list of Data science and Data analyst courses below:
1. Data Science in python by Datacamp
2. Data analyst and Data science course from Udacity
3. IBM Data Science Professional Certificate [coursera]
4. Applied Data Science with Python Specialization [coursera]
5. Data Science Specialization [coursera]
6. Data Science by Harvard [edx.org]

The list may seem too big and you may get confused regarding which ones to take up first. However per my experience all are too good. Start with any 2
from these at the same time and go for it.It will give you an overall idea about data visualization, data inference and modeling, data cleaning and wrangling, R and SAS visualization, plotting and all. Also they cover basics of maths which is needed for machine learning and basic machine Learning concepts along with capstone project.Most important thing here is consistency. This process is time consuming and it takes a lot of patience and hard work. But you have to give your time anyway. You have to find out motivation for that. If you really fail maintaining consistency with the learning, pause for a while, but don’t stop. Stand up and start again.

#3 Machine Learning Maths Foundation:

Time to start our actual machine Learning. Let’s get back to Maths.
We are all aware that computer only understands the language of only 0’s and 1’s.Real world data is too complex and in order to make it useful, we need to apply maths on that complex data. Check Below links.
1. https://www.edx.org/course/essential-math-machine-learning-python
2. https://www.coursera.org/specializations/mathematics-machine-learning
3. Udacity courses on Mathematics.
4. Khan Academy courses on maths.

This link is important:
https://elitedatascience.com/learn-math-for-data-science
Take either Udacity or Coursera courses. From machine learning courses — 3, 5, 6 we are already a little aware of the maths. The idea is to transform maths formulas in to code which will help for data wrangling, understanding important features, relation between features, probabilistic output, data visualization and plotting. And more important testing of model and tuning the accuracy of model and performance.

#4. Machine Learning Foundation:
1. Machine Learning by Stanford University on coursera.
(This is one of the best machine learning course) 
2. Machine Learning Specialization by University of Washington.(I really love this course since they have covered everything in detail and with python hands on.)
3. Advanced Machine Learning Specialization[coursera]
4. Google Machine Learning crash course.

Now Udemy has some beautiful courses on Machine Learning and Data Science and list is as below.

  1. Machine Learning A-Z

2. Python for Data Science and Machine Learning

3. Data Science A-Z™: Real-Life Data Science Exercises Included.
I will recommend above three courses for a newbie. First finish these 3 courses and then jump to Stanford ML course .Another important thing is to follow below blogs and practice ML coding.

#5. Blogs and Site:
1. https://towardsdatascience.com
2. https://www.analyticsvidhya.com
3. https://medium.com
4. https://machinelearningmastery.com

#6. You Tube channels:
1. Sirajology
2. Sentdex
3. Giant Neural Network
4. Machine Learning TV
5. Luis Serrano
6. 3blue1brown

#7. Practise Problem:
1. Kaggle.com
2. Datacamp.com
Now during this process we have to follow your local area meet up on ML and DS. Also need to follow git repos of other people to read their code and make
optimization in code. At the same time publish your work on git account.

#8. Final thoughts:
The above learning path is prepared for a newbie. It takes more than 10 months to learn Data scientist and Machine learning concept free of cost. Strongly agree that this learning process is time consuming and needs lot of passion,courage, patience and hard work. However, at the end we have to follow our own passion and need to make our choices accordingly.

Note: All the content mentioned above is from Internet. I am not promoting or advertising any course or its content, everything is free and available on Internet but in order to build real world projects, we need subscriptions of coursera , edx or udacity, but most learning content is free of cost. In order to buy certificate you need to investment some money as per course fees and policies. I feel instead of investing too much money on any institute for learning these things, it is a good idea to get few real time hands-on with less money with word class universties is better idea .

Special thanks to Coursera, Edx, Datacamp, Udemy,Udacity and all You Tuber and blogger.

Thanks for reading this blog. Happy Learning and Coding.