Number crunchers during this election season

Hyderabad: Until last week, Hyderabad-based Modak Analytics was just another start-up mining and analysing Web data. But all that changed when the 10-member firm revealed to the world what it was really up to inside its “very cramped” office at Kondapur.

Because in just under a year, Modak had built India’s first Big Data-based electoral repository system and that information has helped a “national political party emerge as a big success”.

So, what’s Big Data. The term is used to describe massive amounts of information available online or offline, which grows exponentially every second. In India’s context, we’re talking millions of individuals, what they do, where they live, what faith they belong to and more.

Big Data is defined by three Vs, volume, variety and velocity. What Modak did then, was to bring together data of 81.4 crore Indian voters that’s 18 terabytes of data, which is bigger than the data mining operation ahead of the 2012 US presidential elections. That involved just 19.36 crore people.

Modak, headed by Aarti Joshi and Milind Chitgupakar, didn’t just stop at accumulating data. The “insights” they gained was sold to a political party, which was then able to design a “tailored campaign” targeting select voters by reworking advertisements and by “customising” region-specific polling booth strategies.

Overall, the party was able to determine which area was dominated by what religion, where was the urban crowd most concentrated and which city had the highest number of young voters.
Milind, the chief analytics officer at Modak, cannot reveal the party’s name but admits, “Modak is one of the key for its success. The party benefitted, yes.”

This rather unique merging of Indian politics and high-end tech happened around July of last year.

“We started by analysing electoral data from Andhra Pradesh and found that a major chunk of young voters, between the 18 to 35 years, was simply missing from the electoral list. While I was presenting this find at a Big Data conference in Mumbai, a ‘representative’ from a political party came up to me and discussed certain possibilities. We were excited and in a week, a contract was signed.”

Much of that deal, remains a secret, for reasons obvious. But DC can reveal that the “party” also partly funded Modak’s data mining operation.

After combing the electoral data from AP, Modak moved on to Stage 2: analysing data from other states. “Results were the same. Young voters were missing. It was a pan-India phenomenon,” adds Aarti.

Along the way though, Modak did stumble across extremely unique information. For example, in Andhra Pradesh, the name ‘Srinivas’ is misspelt in at least 600 different ways and over 7 lakh men share the same nickname, Chinna.

“It’s only when we got going with the project that we realised what we had actually gotten ourselves into. I mean there is so much language diversity in India. So, we had to make translation software that could translate 11 Indian languages into English,” adds Aarti. Some of that software, which was used to “teach computers” about India will now be patented by Modak.

For example, to crack the ‘many Srinivases’ problem, the team had to come up with a Fuzzy Logic Matching tool and to co-relate and assimilate such huge and diverse data, Modak had to come up with its more important piece of software, RAPID – ETL (Extraction, Transformation and Loading). “It automated the whole process of mining. We will now apply for a patent,” adds Aarti.

Now that the election is over and their “client” has won, Modak is taking its population-crunching to new levels. “We are in talks with a bank who wants to embed Big Data analytics in its system,” says Milind.

“If our  banks adopt the system, they will rid themselves off a huge problem loan defaulters. If you determine the behavioural pattern and demography of a loan-seeker, banks could benefit.” adds Aarti.

Meanwhile, Aarti has taken Modak to ISB. “It’s worth a case study,” she adds. And Milind has been approached by an international magazine to describe how Modak went about “mapping India” and how a little Hyderabad start-up helped with the world’s biggest election.