I love martial arts. Let me clarify, I love the idea of me doing martial arts. Last year, I was very inspired by this ancient Indian martial art called Kalaripayattu - believed to be the source of all martial arts. So, I took the plunge to start learning the craft of Kalaripayattu. I joined a renowned institute in Bangalore to learn the basics and diligently started attending the classe to do the repetitions required to learn it. It was tough, hard and tiring, yet I persevered initially. But I only lasted for about two months and then gave up. I was not able to sustain the deliberate practice required to continue to learn it. And in some way, my focus was not on mastery, but on this idea of me doing martial arts.
Learning any new craft is a hard and slow process. To do so requires following a very similiar roadmap - seeking inspiration, starting with basics, doing repetitions, conducting deliberate practice and focusing on mastery. I believe, data science is also once such craft that can only be learned in this slow deliberate manner. There is no magic blue pill to learn it.
Now I get questions nearly every week from a beginner on how they could start their data science journey. I am not sure how many really want to do data science and how many are in love with the idea of them doing data science. Still, if they are willing to take the slow, thoughtful approach to learning the data science craft, then here are ten pointers that I would offer from my own experience.
Understand your learning style (shailli - शैली). Are you the hacker kind, who learns best by finding out how things work or do you prefer the structured classroom style learning. Identify how you learn new things.
Find a community of like minded people (sangha - संघ). Most of our learning happens through peers. Build or join a community of peers with whom you can engage, teach and learn. It could be virtual but even better if it is face to face.
Learn the data science process (prakarena - प्रकारेण). Grok the approach to problem solving with data - Frame, Acquire, Refine, Transform, Explore, Model & Insight. And you can only grok by doing it.
Understand the basics (gunas - गुणों). Always investigate why the algorithms and libraries work the way they do. Aim to build a deeper intuition around what is happening in the process.
Practice, Practice, Practice (abhyaas - अभ्यास). Or to simplify just Practice. Repeat the data science process many times and as much as possible, try to write the code, break things and see how to fix it.
Do projects that interest you (ruchi - रूचि). You are more likely to stick through the long arduous journey of doing it. It could be in your work domain or it could be sports, economics, politics, public policy etc.
Share what you do and get feedback (pratikriya - प्रतिक्रिया). Put yourself out there. It may be a small project, a visualisation, an analysis, using a library, anything. Nothing is too small to share.
Build a public profile (pradarshan - प्रदर्शन). It could be a blog, a website, GitHub repos - so that people can see what you are doing and connect if they wish.
Find your own rhythm (laaye - लय). Build and do something every day, week or month. Find your own cadence and start creating.
So take your first steps in building your skills for data science, and plan for a more thoughtful and measured approach to accomplish it. And if you would like a more visual presentation of this essay, go and check out the slides on speakerdeck.
05 July 2017