A hack from Fuzzy Logix makes big data analytics simpler and cheaper

The traditional way to do analytics is to fetch data from a database, store it in separate file servers, create datasets, and use tools for data mining or predictive analysis on the datasets. Then the data is stored back in the database, and reports are generated on the analysis.

This works fine up to a certain level of data, explains Partha Sen, CEO and founder of big data analytics firm Fuzzy Logix. But when the records run into tens of millions and there are thousands of variables involved, as in predicting the chances of somebody getting diabetes, this becomes an inefficient and expensive process. Moving such huge amounts of data from database to dataset and back for each query takes a lot of time.

“When I was working in Bank of America,” says Partha – which is where he worked before founding Fuzzy Logix – “I found that analysts were spending time not so much in analysis but in simply moving the data.”

Funding for R&D and expansion

The US-headquartered Fuzzy Logix, with a development center in Bangalore, India, has a flagship product called DB Lytix which embeds analytics within a database. So a query to the database can produce analytical results directly without the need to first extract a dataset. This not only saves time but also cuts infrastructure and manpower costs.

When data volumes get large, as in the case of healthcare, retail, or financial services, two-thirds of the analytics processing time goes in extracting the data from a data warehouse and storing it in datasets on separate servers for analytics. The “in-database analytics” of Fuzzy Logix can cut that time by half. It also claims to halve the costs of infrastructure and manpower by doing away with the need to maintain the middle tier of analytics servers and the large bandwidth for repeatedly transferring huge amounts of data between the data warehouse and the servers.

Fuzzy Logix today announced a US$5.5 million series A funding from New Science Ventures to build up the R&D for its product and find new customers around the world. “They have a strong customer base, with several Fortune 100 enterprises. During the diligence process, I was blown away by the loyal following from many of these customers using DB Lytix to address complex business problems,” says Vivek Mohindra, general partner of New Science Ventures.

The Fuzzy Logix founder explains the logic of having New Science Ventures as investor. “They bring a wealth of experience in working with Intellectual Property based startups like us,” says Partha Sen, who is an alumnus of Indian Institute of Technology (IIT) Roorkee.

A library of algorithms

Aside from taking up more time, traditional analytics is also limited by the amount of data that can be extracted to a dataset. This constraint is removed by in-database analytics, which is becoming increasingly important with the amounts of data involved in areas like demographic data for marketing insights. It also makes the analytics real-time, unlike a dataset extracted periodically from a data warehouse.

What makes in-database analytics possible is a tool with a built-in library of algorithms and functions to serve the needs of most business users. Building these algorithms and models was a hobby for Partha from the time he started working for Indian IT giant Tata Consultancy Services in the mid-nineties and persisted even when he later became senior vice-president for Bank of America.

Multi-billion-dollar market