Are you a self-made data scientist? How did YOU do it? by Sudalai Rajkumar S

Answer by Sudalai Rajkumar S:

Here is one question which I think am qualified to answer for.!

I am a Mechanical Engineering graduate who had no prior knowledge about data science or for that matter even coding when I left my college six years back. Now I am working as a Lead Data Scientist in a reputed firm and also one of the top 25 Kaggle Data Scientists in the world.

Though I do not have formal background in CS or Statistics or Maths, I have a passion for crunching numbers and finding patterns right from my school days. I think anyone with a good passion for patterns and numbers coupled with right amount of hard work can become a self-made data scientist. Here is my path :

MOOC Courses:This played a major role and is the first place in my learning path. Courses which helped me understand the basics concepts are

- Introduction to Statistics by Edx – This is very good introductory course in Statistics which taught me the basic concepts
- Machine learning course in Coursera – A very famous course by Andrew N G which most people are aware of
- Analytics Edge course in Edx – This is again a very good course with a lot of practical examples
- Statistical Learning by Standford Online – This is again a very good course by which teaches the concepts of predictive modeling in detail with R codes. The curriculum of the course closely follows this book
Some other nice online courses which I came across are

- Data Science by Harvard Extension – This is a very good course for people wanting to learn the concepts using python.
- Data Science and Engineering using Apache Spark by Edx – This is a very useful course for people starting with big data analytics
- Learning from Data by CalTech – This covers the basic concepts of machine learning
- Neural Networks for Machine Learning by Coursera – Interested in knowing about the new boy (Deep Learning) in town. This course is the perfect place for that taught by none other than Geoff Hinton himself.
Once I get a fair understanding of the DS concepts from these courses, I was itching to use them somewhere. I was looking for options to test these theoretical skills. That is when I came across DS / ML competitions.

DS / ML Competitions:I came to know about Kaggle when I was searching for datasets to apply my learnings. I thought that I can ace the competitions easily since I have a good understanding of basic concepts. Poor me was not aware that hands-on is a different ball game from theory.

I started doing competitions on Kaggle but ended up at the bottom half of the table inspite of all the hard works. So once the competitions were over, I started looking at how others solved the problems from Kaggle Forums and blog. This is one important place where most of my learning took / taking place.

It also helped me hone my structured thinking on approaching the DS problems. It also helped me work on different real world datasets from different domains, each one challenging in its own way. When working deeper on these problems, I got new learnings every time and helped me improve myself further.

Doing Kaggle competitions at the first go might be daunting these days since the competition levels are quite high. So one can try to work on data science problems in other platforms like Analytics Vidhya Hackathons, Crowdanalytix, Driven Data etc before trying out on Kaggle to gain some confidence.

Other Sources:Apart from MOOCs and DS competitions, two important sources that helped me with my learning and understanding of this space are

I follow these two blogs to update my knowledge and to keep up myself to the advancements in the field.

Other resources which I found to be helpful are

- Data Science Central
- WildML blog
- Analytics India Magazine (To understand the happenings in India)
- MLWave blog
- FastML blog
Hope this helps other budding self-made data scientists.!

Are you a self-made data scientist? How did YOU do it?