The role of Mathematics and Statistics to become a Data Scientist
My Short Story
Data Science is Hard. And, there is so much to learn.
When I first started with Data Science I was totally baffled by the Mathematics behind.
So What did I do?
This post is my story of how I started with Data Science. It is a true story.
A Brief Background
I am a Mechanical Engineer.
No CS Degree. No formal Math/Stat education.
Got a business analyst job in 2010. For the first three years did a lot of grunt reporting work including Excel/SQL, Spotfire.
Then the opportunity presented itself and my manager asked me to his office. It was the year 2013:
Manager: I want you to learn some Python.
Me: But I have never done programming in my life. Why now?
Manager: Have you heard about Data Science? There is something called the Random Forest Algorithm that I heard about. Since you are the only non-developer around I thought you might be able to dig in(Business Read: Since you are the only one with no work on his hands, why don’t you do something)
So I went to Google.
Read what RandomForest is used for. Just understood a little about the classification problem statement.
Read a blog from someone who had implemented it in Python. Copy Pasted his code. Ran on my local machine. And Voila!!!
We used that small piece of code to run on a classification dataset for that company and included it in a product. My manager was impressed.
At that point in time, I really didn’t know any data science. I didn’t know about Entropy, decision trees, and Cross-validation. Hell, I didn’t even understand Linear Regression.
To tell you the truth the model I created might have overfitted as hell. But I just know that it was a start. And I consider that start very important.
From then on I started trying many models to do the same task.
I started up with Kaggle. And when I was not able to compete I started reading more.
I took many open MOOCs. I learned about new things.
I was open to learning new things. I was open to understanding new things.
And then one day I realized that I was able to understand the math behind if I was willing to put effort.
I started up with the breadth and eventually got into the depth of things.
So yes, get your start. Get your hands dirty.
Run the algorithm/s. Try to improve them by playing with the parameters and reading up about them.
I am sure in the process you will learn a lot about the inner workings of the algorithms.
And maybe then someday you will be motivated enough to understand them fully.