How I learned Data Science without spending a penny.

Data Science is a vast field and learning it could seem like a daunting task. There are so many pre-requisites as well. To cover such a vast field you need a well-organized plan. And this is precisely why a lot of people end up spending a large sum of money to join learning programs. Remember, they charge you for the organized content. Some programs even claim guaranteed placements, but mostly there’s a tiny little star (ie. terms and conditions apply).

I have always been a self-learner (I don’t know why I just can’t focus in a classroom 😅) and so I always try to promote self-learning. So here I am going to help you start learning data science by yourself. I will also suggest some of the best freely available resources for learning the respective topics.

So let’s start with the pre-requisites first:

1. Linear Algebra: You need to learn a bit about vectors and matrices, and operations like dot products and matrix multiplication, etc.

• 3Blue1Brown’s playlist “Essence of Linear Algebra” is an excellent resource for developing an intuition for Linear Algebra.
• You can also try the Mathematics for Machine Learning Specialization Course 1: Linear Algebra.

2. Calculus: Now, you don’t need a lot of depth in calculus to start learning data science but a decent knowledge of the basics is required to help you understand the Backpropagation Algorithm which is one of the main ingredients for Machine Learning.

• Again 3Blue1Brown’s playlist “Essence of Calculus” is an excellent resource for developing an intuition for Calculus.
• You can also try the Mathematics for Machine Learning Specialization Course 2: Multivariate Calculus.

3. Statistics: Stats is probably the most important prerequisite for data science. With the abstracted tools, you might be able to do just fine without much linear algebra and calculus (although I wouldn't suggest it), you won’t be able to understand the results if you don’t understand statistics.

• StatQuest’s playlist: Statistics Fundamentals is a great resource to learn statistics.

4. Python: Python is the most common language for Data Science and it’s easy to learn. Although, you don’t just need to learn python but also some of the common libraries like NumPy, Pandas, Matplotlib, Seaborn, etc.

• Krish Naik’s python playlist cover’s everything from installing Anaconda to the libraries I mentioned above.
• Also, a bit of practice on Hackerrank wouldn’t hurt, especially if you don’t already have a programming background.

So that’s all the pre-requisites you need. You’re all set to start with the good stuff. Now, there’s a lot of theory you might have to go through but as long as you’ve learned the pre-requisites well it shouldn’t be that hard.

1. Machine Learning: I made the mistake of directly jumping to deep learning and lost interest soon(the first time I tried learning). Don’t make that mistake. Start simple and slowly build your base, once the basics are clear you can easily build upon that. Don’t forget classical Machine learning is still the go-to option at least for structured data.

• You might have heard of the Machine Learning course by Andrew Ng on Coursera.
• There’s also Stanford's CS229 Course on YouTube which is an in-depth version of the above course.

2. Deep Learning: The problem with classical machine learning is that you have to do the feature extraction manually. However, that’s not very intuitive, especially in the case of unstructured data. There comes deep learning which is an end-to-end solution that learns to do the feature extraction as well.

• Again, Andrew NG’s Deep Learning Specialization course on Coursera is one of the best there is.
• There’s also Stanford's CS230 Course on YouTube which is an in-depth version of the above course.

With this, you’re all equipped with enough knowledge in data science to start working on projects. Now, there are a lot of subfields like Computer Vision, Natural Language Processing, etc but those will require a separate blog. This one’s already long enough and I don’t want to scare you away from this field😅

In the end here are a few tips to help optimize your learning process:

1. Watch videos at 2x speed. Don’t directly jump to 2x of course, start with 1.25x and slowly build up your speed.
2. Use the audit option on Coursera to learn for free if you don’t want the certificate. If you want the certificate as well, you can apply for financial aid if you can’t afford it.
3. Read books: There are some really good books which you can go through. DL by Ian Goodfellow, Pattern Recognition and ML by Christopher M Bishop, ML Engineering by Andriy Burkov, and Approaching (Almost) Any Machine Learning Problem by Abhishek Thakur to name a few.
4. Go through Kaggle kernels and discussion sections. You will find some good content there.
5. Practice, practice, practice. There’s so much theory that it can feel mundane at times. Use that opportunity to apply the things that you have learned so far and work on projects.

That’s all, good luck and keep learning.