What's the best thing about being a data scientist

Movies to Mahabharata, he's analysed them all

S Anand's passion for computing drove his Chemical Engineering professors to distraction at IIT Madras, and when he moved on to IIM Bangalore, after a brief stint at IBM India, he was nicknamed "Stud" and alternatively, "Prof" — which as any alumnus will tell you — is one of the highest forms of praise that can be given to a batchmate. After graduating from IIM B, Anand moved on to jobs at the Boston Consulting Group and Infosys Consulting. In 2012, he was infected by the startup virus, and became chief data scientist at Gramener, which specializes in data visualization. Anand lives in Bangalore, where he uses text analytics to uncover the fact that it is Yudhishthira who is at the centre of Mahabaratha's plot, and not Arjuna, as one would imagine. Oh, and he also uses data science to understand his mobile bill.

On becoming a data scientist ...

It was an accidental mix of three of my personal interests — programming, statistics and design. It wasn't until 2009 that the term "data science" was coined. Until then, I knew I wanted a career that combined all three of my passions. Once I knew it had a name, the decision was obvious.

What's the kind of work you do at Gramener?

We're short circuiting the gap between gathering data and understanding data. We're work with all kinds of data. We work with aircraft flight paths, call data, court case results, fielding performances, poll forecasts, TV serial transcripts, even sex survey responses.

But across these domains and types of data, there are common patterns of problems. Which are the biggest problem areas? What factors drive performance? What is the expected outcome? Fortunately, we have also discovered common patterns of analysis and visuals that answer these. Our work involves applying these patterns to any and every dataset.

Best thing about your job ...

Stumbling on strange patterns that tell a story. For example, when we discovered that 88 people at Modakkuruchi, TN, stood for elections but didn't even vote for themselves, it was a shock. Or when we found that children born in August score much more than children born in July. Or that people taking meter readings don't always visit the premises and cook up numbers that are typically multiples of 10.

There are usually several insights hidden in data. For us, it's like exploring Alibaba's cave of treasures.

What tools do you use?

We use Python (a programming language) for most of our analysis. It's easy to learn, and easy to explore data with. R and Julia are quite popular as well. But Excel remains the primary method for data exploration in most organisations, and is quite effective.

However, the tool is often less important than the scientist. A good scientist with a paper and pen will beat a moderate scientist with the best tool available today.

How do you use data in daily life?

Several of my day-to-day decisions are data-driven.

What to eat? I scrape price and calorie information to pick cheap energy-rich foods.

What to watch? I visualise the top-rated popular movies by genre from IMDb.

When to book a flight? I review flight price trends, and typically book exactly 8 days before the flight.

In many ways, my becoming a data scientist was driven by these inclinations. It's not because I'm a data scientist that I do this. It's because I have always done this that I'm a data scientist today.

(As told to Narayanan Krishnaswami)

Numbers have a story to tell

Somewhat like endurance biking, data science is about staying power and enjoying the ride. Little wonder that Raam Nayakar, who is senior manager of marketing analytics and strategy at fashion e-retailer Myntra, follows one as a passion and the other as a profession. Sometimes the twain do meet when the 37-year-old uses his data orientation to find petrol stations and eating joints on remote biking routes.

Best thing about the job

It's a dynamic field. Finding new ways to analyse data keeps me on my toes. For instance, I recently learnt Python, an open-source analysis system in demand these days.

Data analysis sometimes results in counter-intuitive insights, which is very exciting for us. For example, we once told an American wholesale retailer's credit card program to invert their loyalty rewards system and reward low spenders instead of the higher ones. We had identified the highest spenders as mostly re-sellers who did not need to be incentivized. This saved the company millions of dollars.Sexiest job of the century? Two data scientists tell us why

What makes a good data scientist?

A data scientist has to have a business context. We should be able to look at numbers, understand what they are telling us, and how that message can be converted into an actionable form. A good data scientist needs to have an eye for detail as well as the big picture.

A typical day at work

One day I am trying to find which TV channels drive better traffic for Myntra; another day I am engaged in text mining to rank social media responses for a contest we ran. It's the diversity of business problems that keeps my work interesting.

Applying data science in daily life

When I came back to Bangalore after a few years of living abroad, I was overwhelmed by the traffic. It would take me an hour and fifteen minutes to reach the Hosur Road office from my residence in JP Nagar. In the first month, I tried multiple routes at different timings, assessed areas of traffic congestion, and road conditions. In other words, I analysed the traffic data to construct a route that now takes me less than 45 minutes.

Being a data oriented person, I don't miss out on details. My skills help me optimize time and effort even when I am not working. For example, if you want to get good properties to stay during your trips abroad — find out from which country do most of the tourists come from. Then make your booking from a local travel site in that country. I got two nights in a private villa in Bali when I booked from an Australian tour agent site. At the time, Australia contributed over 25% of tourists to Bali. Now, China has overtaken Australia.

Is it going to stay a hot job?

Many startups are now offering data science incubators and boot camps to train an army of data miners. Business analytics training institutes are offering courses, as are some of the IIMs. Big data is going to get even bigger.

source: techgig