01 Jul 2015 20:46 IST

‘Big data analytics isn’t only for scientific problems’

The capability of organisations to store data, and micro-level data, for longer periods has gone up.   -  shutterstock

It can also help predict consumer behaviour, says S Ganesh, CEO & Managing Director, Dun & Bradstreet Technologies

With consumer-facing industries using data analytics to predict consumer behaviour and spending, big data analytics has moved out of the scientific realm and has emerged as a new profession for number-crunchers. In this interview, S Ganesh, CEO & Managing Director, Dun & Bradstreet Technologies and Data Services Pvt Ltd, says it isn’t just those with a knack for numbers who can become data analysts. Those with a spatial bent of mind, from architects to marketers and those with a humanities background can enter the field too. Excerpts from the interview:

Companies have always used statistics and they’ve been number-crunching. But, of late, data analytics has become quite the currency of usage and has emerged as a fairly new profession. So why has Big Data suddenly become part of the lexicon?

Two primary trends have essentially led to the use of statistical analysis for prediction becoming a widespread phenomenon across multiple organisations. The first is that the cost of data or the cost of data storage has decreased exponentially due to Moore’s Law .

The cost, 15-20 years ago, of 1MB of data has now translated to the cost of 1TB of data. This has caused an explosion of data availability in organisations across the globe. Also, the capability of organisations to store data, and micro-level data, for longer periods has gone up.

The second, a corresponding phenomenon, is the dramatic explosion of technology in crunching that data. Today we have microprocessors in our pocket which, 30 years ago, would have filled up a whole building. With these two phenomena, statistical techniques for prediction have gained wide currency and more and more organisations are storing and mining data for relevant insights in a variety of areas.

This could cover buyer behaviour, purchasing propensity, timing of propensity, credit behaviour, payment behaviour and, in many cases, even predicting things like travel behaviour and movement behaviour. Now with the explosion of GPS and other geospatial kinds of programmes which are always on, prediction of even localised kind of analytics is possible, where you can predict that if a person is in X place, he will behave in Y manner.

Would you say that since data analytics has come out of the scientific realm and is now consumer-led, it has led to Big Data analytics gaining currency?

Absolutely! Today, analytics is no longer only for esoteric scientific problems. It is, and will be more and more well-embedded into everyday behaviour of human beings. To that extent, the industry has to offer up solutions that are easy to understand, quick to use, and of relevance at every given point of time.

Now that there’s a huge interest in data analytics even in B-schools, typically what does the industry look for? Do you look for engineers, or do you look for business school students, or a combination of both? What does one need to study to get into the data analytics space?

Definitely, to be good in maths is a prerequisite. Even if not very good, at least you have to like it because you’ll always be talking about numbers. But there are multiple threads and strands in analytics. The first thread is that of pure data scientists or data analytics professionals who are given certain problems. And they give their reply in terms of the propensity of a certain outcome to happen.

For example, if you’re given the delinquency behaviour of a portfolio or repayment behaviour of a portfolio, they will predict how it will behave in the future, given certain variables. That is the first and that is a very important job. But that's not the only one. There are spaces for a lot more people.

There are lot of marketing analytics which have come up. For this, you not only need to know mathematics or data, but also a lot about customer behaviour, because you can have false positives. Sometimes, data can throw you very surprising things. So, for marketing analytics you need people with an analytical bent of mind. The ability to ask the right questions, and to understand what the desired outcomes are. For example, how do I increase revenue from a customer can be a good question. How do I become relevant to a customer can be a second.

Then the third stream, which is not so much understood, is that there is a lot of space for arts and artistic studentsin data analytics. We are talking of not only simple correlations but also pattern correlations, wherein we can say a certain person falls into a certain kind of behaviour pattern, and certain people of similar behaviour do certain kinds of things.

Now, pattern recognition is less of a mathematical science and more an artistic science. And, there are people with good knowledge of humanities, of geospatial knowledge, of 3D knowledge. Good architects would be very good in analytics. So, I would personally say this is a field open to anybody who is interested in predicting human behaviour, to anyone interested in predicting the behaviour of consumers, businesses, entities, even inanimate objects like automobiles.

When you recruit, what do you look for in a candidate?

There are two kinds of people we recruit. We do independent, third-party data projects for customers and banks across the globe. For this, we primarily look at mathematics graduates, scientists, statistical graduates, and technology graduates with knowledge of analytical tools, techniques, problem-solving ability, and so on. Other than this, we also do predictive algorithms, such as predicting scores of individual repayment behaviour in an uncertain economy.

For that, we need statisticians, economists, people who have a good understanding of economic behaviour and trends that they can incorporate into real analytics. For this, we hire people with a background in mathematics, technology, statistics and economics. We train them in analytics, in problem solving ability and capacity, and get them trained in certain tools, such as SAS, which help in pattern recognition.

Recommended for you