Sramana Mitra: Can you please position this for me in the context of your competitive landscape? We’ve tons of Big Data companies. We’ve seen this problem from many different angles. I’d like you to get granular in positioning your product based on exactly what you do.
Radhika Subramanian: If you have data, the three steps in analyzing around data are collect, analyze, and report. Those are the three steps. There are tons and tons of companies that are in the collection and storage space. It’s very mature and growing fast. Tons of innovation is happening there. Emcien has nothing to do with that. Let’s now go to the reporting side. Reporting was born because once you started taking data and shoving it to databases, businesses really struggle to get the data out. That’s really how BI was born. That’s reporting.
Let’s come to the middle part, which is analysis. Analysis is beyond reporting. Analysis is being able to do more with that data. That’s the space that Emcien plays in. Now, let’s take the analysis and let’s try to drill down into what analysis means. Historically, we have had very low volume Big Data. The reason the entire field of Statistics was born is because we had very small samples of data and we would extrapolate it into population. The entire field of Statistics works with small data. The second thing is we’ve always had numerical data. The kinds of data had been very limited. As the Internet started to grow, the language of technology is data – all kinds of data. Numerical is a very small part of it. The higher field of Statistics is now facing this huge amount of data that it’s not really equipped to handle. The science hasn’t been discovered to actually analyze it.
The next set of method is around data mining. Tons of stuff gets underneath it but a lot of them are search and query based. In fact, all of Statistics is query based because a hypothesis is a query. You’re really testing if this is true. As data gets big, the number of questions you have to ask of data to get insight is very laborious. It’s a very difficult way of analyzing data. If you look at Hadoop, Hadoop is essentially taking the query method and saying, “What if I could speed it up?” Then you have databases that have been specifically designed to speed up the querying process. But still, it’s like trying to take a bicycle and saying, “How can I fly this to the moon?” as opposed to saying, “Maybe what I really need is a rocket ship.” It ends with what we invented or created completely by chance.