Sangeeta Apr 16, 2014 No Comments
Beginning this month, IVY Blog shall feature a start-up that brings in a fresh wave of innovation or unusual perspective to the field of analytics.
With elections dominating conversations, we decided to feature ZT Analytics as “the start-up of the month” for the innovative web platform, ‘A Social View of India Elections 2014’ designed as an exclusive election analysis portal.
ZT Analytics is barely four months old. Yet, it did not deter the brains behind the start-up, three Analytics enthusiasts, Madhu Jalan, Raghuram Lanka and Reva Maheshwari, from leveraging their varied experience to make a bold statement this election.
We asked co-founder Madhu Jalan about “what prompted them to launch this initiative when news channels are already churning out social media analysis reports by the second?”
There is a huge amount of information being published about the current elections in the news and through social media. For the first time social media is giving us a platform to get the opinions of the common man and especially the youth real time.
Usually general issues stay in the news for a very short period of time. The elections have given us an opportunity to analyze data over a period of almost 4-6 months – before, during and after the elections.
This rich dataset of opinions, conversations and ideas are available for anyone to use. It is a dream come true for any analytics company that is focusing on text mining and social media analytics. We couldn’t possibly pass up the opportunity to analyze the elections regardless of the number of reports being churned by news channels.
What are the analytics tools used? Any particular vendor software / tool used in the model?
No, we haven’t used any particular vendor software for our sentiment modeling. We have built our own set of proprietary tools. These tools are written in Python and use some of the existing libraries available.
Buzzwords keep changing and evolving during elections. What is the text mining algorithm used in your model to keep up with the flow? How was it derived?
We used our own proprietary modifications to common algorithms.
Besides elections, where else does your firm see the application of sentiment analysis and text mining in India?
Many different industries can use sentiment analysis and text mining. A lot of people are trying to follow conversations around stocks to see if they can form an opinion on ‘buy, sell or hold’. Similarly, the airlines industry is following social media to understand customer grievances. Companies are also using social media to track their brand image, perception of products old and new, and customer service. Given that a company, on average, generates ten times more unstructured data than structured data, text mining should be of use to company looking to analyze their data to make business decisions.
What is the ZT Analytics giveaway for students of analytics? Any inspirational tips that you would like to share?
Being a student is a wonderful time to experiment. There are a host of open source tools today as well as open sources of data which make it relatively easy for you to start experimenting. This is the only way for you to learn and to understand what works and what does not.
The other thing that you should not ignore is domain knowledge. Without understanding your domain and the data, you aren’t going to be able to come up with meaningful insights. So take the time to understand the domain, the data and what would be considered a ‘meaningful insight’ before jumping in and experimenting.
Would you like to comment on the use of social media analytics in the ongoing 2014 elections?
For the first time in Indian politics we are seeing extensive use of social media by parties and candidates. It is a great place to understand public sentiment around the elections, candidates, parties and issues. Social media is also a great channel to reach out to the youth, especially in India, where more than 150 million of the 815 million voters are between the ages of 18 and 23 and will qualify to vote for the first time.
However there is a lot of junk on social media. No one has to take accountability for what they put out on Twitter. In our algorithms we compensate and validate for this by using online news media for our analysis in addition to social media.
Another thing to think about is that Twitter is still mostly an urban phenomena. It reflects the opinions of a small group of Indians. It is extremely important to understand the demographics of the population on social media when making statements about its reach and influence.
Madhu Jalan, is a senior, entrepreneurial professional with diverse experience in new business development and strategy. With a Matheatics background followed by M.B.A., M.S. and a recent certification in Analytics, she is all set to bring in her diverse experience to Analytics, supported by her team-mates.