Best Resources to learn AI, Machine Learning & Data Science

A lot of people have in mind how to become a Data Scientists or Machine Learning Engineer. Hence I have tried to make a compilation of some of the best resources one should learn from basics to advance with a sequence of courses & blogs. I have also provided many amazing additional resources and tips which I wish I had known when I started out. I have got for you almost everything that you need to be Data Scientist or ML Engineer.

Why AI/ Machine Learning?

In this era of digital revolutions data is floating in trillions and billions, the advancement in Machine Learning has proved to be the savior. The volume and the scale at which we people are producing data is humungous, but to tap data and convert them into meaningful information we need to feed them into a computer and analyze it at length and breadth which is where ML comes into the picture. Undoubtedly it’s going to shape the fate of how companies and countries will take decisions. Our computers are used for analyzing petabytes of information in seconds, so ML techniques, when enhanced with right kind of data, can give a distinct advantage for sectors like Science, Healthcare, Banking & Finance, Manufacturing and many more. It has endless applications.

Credits — Harvard Business School

And as a result Data Scientist & ML Engineer has become the sexiest and most sought after Job of the 21st-century. AI, Machine Learning, Deep Learning, Data Science are the buzzwords all around. Two years before I started with my journey to learn this superpower AI with which you can solve nearly every problem and has unlimited applicability. I was very fascinated by this field but was feared to dive into it as I was not that great in maths. But I decided to give up my math phobia and thought of giving it a shot. See If you’re really passionate about the field there are a pool of free resources available, you just need to have that enthusiasm and motivation.

Disclaimer: I am still learning as mastering AI/ML as it requires a lot of time, whatever I have mentioned here is the path I would have taken If I were to learn ML/ AI today and the things I wished I would have know before. I will be regularly updating the blog as I get more better resources.

Recommended Videos for Inspiration into AI

Will recommend going through the following videos, it will give you what is AI exactly and provide you with the inspiration and range of possibilities of AI. As Online Learning has a very high dropout rate and most of the people don’t even complete 10% of the course they take up.

So a Strong WHY is very very crucial.
Are you just after the hype or in it for handsome pay?
Or
Do you want to build things, apply them to solve endless problems and to make a difference?
Make sure you have it by the end of seeing these videos it will keep you going as you might lose motivation down the way.

What is AI exactly? — Coldfusion
A.I. is Progressing Faster Than You Think! — Coldfusion
“I Am AI” — NVIDIA
The Rise of AI — Bloomberg
Real Difference between AI/ ML/ DS/ DL — Applied AI Course
Why Deep Learning Now? — Coldfusion
How to Study Machine Learning — Siraj Raval

What it takes to be a Data Scientist

Don’t get daunted by the above image, its true that a Data Scientist must have above skills but remember Data Scientist is a very senior role and it requires years of experience and continuous learning to reach to this level of expertise.

But its good to have in mind that you need to reach to this level at some point of time and look it as a challenge for yourself.

Learning Approach

Take the Breadth First Approach. Don’t try to go too deep into one topic in ML/ AI as you will enter into a rabbit hole. Better is to take a flavour everything from a basic perspective and then revisit it again and again in iterations with some advance concepts each time. Some people taking the Depth First Approach lose their motivation and end up giving up.

Will suggest you doing the Beginner Courses first and then go for Advanced courses given below. Will recommend you to following the sequence. Feel free to skip any section if already done you can anytime come back later.

There is a huge pool of courses out there! Confused about how to go about it?
Here are some of the best courses in a sequence according to me.

Python

Beginner
Introduction to Python Programming — Udacity
Introduction to Python — Kaggle Learn
Pandas — (Kaggle Learn)

Advanced
Python for Developers — Jupyter Notebook Collection

Directly go for Kaggle Learn if you have previous programming experience.

Maths

Beginner
Linear Algebra — 3Blue1Brown
Probability & Statistics — Khan Academy
Calculus — 3Blue1Brown

3Blue1Brown has mind-boggling visualisation and the way he explains the difficult concepts is such you will think you would have invented Calculus yourself. Don’t get stuck completing Maths at Khan Academy after covering the basics start with ML and come over it again and again.

Advanced
Computational Linear Algebra — Fast.ai
Multi-variate Calculus — Khan Academy

Rachel Thomas is an amazing teacher and explains Linear Algebra with practical examples and code in Jupyter notebook and solving some simple problem a fun way to learn Linear Algebra.

Machine Learning

Beginner

Either one
1. Applied AI Course

Personally I found this as one the best course out there for ML/DL. It has all the things available at one place that you need to master this skill. It teaches from scratch Python all the way to Deep Learning with various amazing cases studies. Shrikant Sir has an infectious enthusiasm with some of the best practices in the industry he keeps you going through learning ML like its child’s play. It teaches all the algorithms with all the rigour of maths needed without overwhelming the students. And provides a good intuitive understanding of the concepts and geometry. Thought it is a paid course but must take if you can afford it. Though I maybe biased on this one as I personally took this course but I really found it amazing hence I highly recommend it.

OR

2. Intro to ML — Udacity / Machine Learning A-Z — Udemy

& Machine Learning — Coursera(Andrew Ng)

I feel both Udacity and Udemy are really good for newbies but it just scratches the surface and don’t go much deep into the topics and hence I recommend after completing that you must go for Andrew Ng ML course at Coursera as he explains the all the maths behind every algorithm quite well.

Advanced

Introduction to Machine Learning for Coders — Fast.ai
Machine Learning Kernels — Kaggle Learn
Machine Learning Explainability — Kaggle Learn

Fast.ai is an amazing place to learn, you will get to learn state of the art techniques with fast.ai own made up library. It takes a top to bottom approach and make you do a lot of coding so that you are able to understand each and every detail. They believe if can code ML you know ML and that’s quite true in my opinion. As 90% of people won’t know to code a Random Forest. If you are really into Kaggle this one is must for you. As Jeremy Howard had been a Kaggle Grandmaster and shares his 25 years of experience at industry. Love this course due to its practicality. Kaggle Learn is good for Revising the concepts. ML Explainablitiy is crucial as you will be able extract human understandable insights from any Machine Learning model.

Make sure for each technique you learn the following :

1. Theory with Geometric Intuition
2. Maths behind it (Please do not skip it)
3. Assumptions of the Algorithm
4. Best & Worst Cases
5. Interpretability
6. How to Code from scratch
7. Variations of the Algorithm
8. Limitations

Data Analysis

Beginner

Data Analysis Pipeline — edX(Georgia Tech)
SQL — (Kaggle Learn)
Data Visualisation — (Kaggle Learn)
EDA Kernels — (Kaggle Kernels Grandmaster SRK’s Kernels)

Its very important for a Data Scientist to understand the Data is the king. Without data there is no ML and concept of Garbage In-Garbage Out. A good structured data can outperform the complex algorithms. So you need to understand the whole data analysis pipeline which involves collection, preprocessing, storage, analysis, and interactive visualization of data. The course by edX on this topic is really amazing. Also you need to know SQL as most of the data in industry is tabular and often stored in SQL. I would highly recommend you going through the highly rated Kernels on Kaggle by Grandmasters

Honorable Mentions

1. Detailed Resouce List on Github
2. School of AI
3. Datacamp
4. DataQuest

Practice

Practice! Practice! Practice! There is no replacement for practice. You might do hundreds of courses online it won't be of any use if you don’t practice. You will get to know the boundary cases, failure cases and even which sort algorithms work best for a certain type of data. Do you know that most Data Scientists are only theorists and rarely get a chance to practice before being employed in the real world. Gaining this experience is very crucial for a successful ML Engineer / Data Scientist.

1. Kaggle
2. Analytics Vidhya
3. CrowdAnalytix

Please read this in order to understand why Kaggle is so important.

Projects

Solve real-world problems which matter most to you and you can take inspiration from the following resources. Do document your project well on GitHub and explaining in detail about each everything. And don’t forget to write the procedure to replicate your code in someone else’s system as some of the recruiters may want to see whether you really have done something you claim

1. Awesome Project Ideas
2. Kaggle Dataset with Tags(Pick a dataset which suits your interest and work on it and build something useful)

Other Resources

1. Papers with Code (All the latest research papers in ML with code)
2. ArXiv Sanity (Top Current Research Papers)

Some More Useful Videos

1. Data Science/Machine Learning Project Life cycle
2. How to get ML Internship
3. Best General Videos Playlist — Siraj Raval

Keeping up with the field

Remember learning AI is a continuous process as its one the fast-evolving fields so make sure keep up with the field on a daily basis. There are multiple resources viz.

Youtube

You must follow Siraj Raval, he one the Evangelist of this field and is doing lot for the community with his videos and SchoolofAI. It good you follow them and learn from them as they break down the complex concepts very well.

Must subscribe to this Youtubers
1. Siraj Raval
2. Arxiv Insight
3. Sentdex
4. Two Minute Papers
5. Deep Lizard

Medium Articles/ Blogs

Read as much as you can medium blogs and some bloggers in ML community write very brilliant articles its strengthens your concepts and you see this of from other perspectives and gain more deeper insights of the concepts

1. Analytics Vidhya Blogs
2. Towards Data Science
3. Sebastian Ruder
4. Colah

Twitter

It may be strange for few but twitter is where the you’ll get the latest whereabout of ML/AI. ML/AI community is pretty amazing and technical.

@jeremyphoward — Deep learning researcher & educator. Founder: fast.ai; Faculty: USF & Singularity University; Previously President: Kaggle
@karpathy — Director of AI at Tesla. Previously a Research Scientist at OpenAI, and CS Ph.D. student at Stanford
@ch402 — ML Researcher at Google, AI Blogger
@twimlai — Brings you the week’s most interesting and important stories from the world of ML and AI.
@ML_Review — Tweets about most influentials recent papers, lectures and projects on Machine Learning.
@stanfordnlp — Computational Linguistics — Natural Language — Machine Learning — Deep Learning. And misc technology from Silicon Valley.
@zacharylipton — Assistant professor — @carnegiemellon (Jan 2018), mad scientist @awscloud, Ph.D. ABD @UCSD
@johnmyleswhite — Manages the NYC branch of Facebook’s Core Data Science team.
@chrisalbon — Data Scientist. Authoring Python Machine Learning Cookbook (O’Reilly, Forthcoming).

Community

1. Wizards (Slack channel by Siraj Raval)
2. Analytics Vidhya Community(Slack Channel)
3. Reddit (/MachineLearning)
4. Kaggle Discussions Forums
5. Fast.ai Forums

References

1. https://www.kaggle.com/sudalairajkumar/where-do-people-learn-ml-ds
2. https://github.com/llSourcell
3. https://appliedaicourse.com

Further any suggestions are much welcomed. You can share any resource which I may have missed out as it can benefit others in the community.

Thank you for the read. I hope that you have enjoyed the article and gained a lot from it. If you really liked it, please hold the clap button and share it with your friends. You can reach out to me on LinkedIn. If you have any questions, feel free to ask them. Stay curious! Happy Learning!