How to study for data science?

Introduction

This article is to help you guys how to jump from web development to data science. I faced a lot of problems when I first took dive into this field. Here, I will list all the resources and how you should study them.

Before going any further I will bring it to your account that being a developer it will be a lot easier for you to implement papers or any project.

Why am I writing this article?

I got an intership at Indian Space Research Organisation. I was introduced to data sciene then and there only. Realizing its value I decided to dive into the field. I faced many difficulties when applying or working with any data set as in how and which algorithm to use. This was mainly because I lacked a good intuition of math behind data science. I learned many frameworks but all of them rendered useless until I found how to approach the subject properly.

Who is this article for?

Any one who wants to study data science and has done some coding for atleast few months.

Study guide

Before going to read or understand any paper, you should be able to understand the language of these research papers. I have spent a lot of time implementing a paper because I never fully understood what mathematical jargon was written in the paper. So for that, I recommend you getting your math a brush up. Here how you should study. All of the courses I state will be either free or free to inspect.

Prerequisites

1. Essence of Linear Algebra by 3blue1brown: This course is a very interesting free course, which will give you a clear understanding of linear algebra. There are numerous animations in this course which will give you intuition about these basic building blocks of data science. And you cannot forget the music, which will engage you to watch his videos till the end.

2. Stat 110x on edX: This course does not only teaches you probability and statistics but how to study math in general. It lays emphasis on importance of story and how thinking math solution through a story will clear you understanding.

3. Multivariate Calculus on Khan Academy: This course is taught by Grant Sanderson who is creator of 3blue1brown. I never knew how important and deep multivariate calculus is until I took this course.

Courses

1. Machine Learning: This course has always been first step for any student learning data science. There is a reason behind this as course is beginner friendly and builds intuition by skipping many gory details.

2. CS7015: This course is offered ny NPTEL by IIT Madras. This course relies heavily on mathematics. This course will build your intuition behind many deep learning concepts. Once you brushed up your math, this course will be very enjoyable. I know there are lot of course like deeplearning.ai specialization but this course covers many details that are otherwise skipped in many other courses.

How to study for these courses?

1. Make notes after you watch a concept. I will recommend you to use OneNote/EverNote. Try to explain what you have learnt and record it so that you don't have to watch the whole video while revising the concept. You can take a look at my notes how I have made my notes here.

1. Learn numPy and scikit-learn. These will help you implement the concept you learn on the way. Believe me, this will really help you. Look at this notebook, I made this when I learned about Jacobians.

1. Read, understand and if possible implement the research papers you read. Use app like wandb.