PUNE, INDIA:  Mukund Deshpande, head - BI and Analytics Practice, Persistent Systems has spent eight years of action in BI that also covers heuristic driven algorithms, classification and prediction algorithms after PhD in Computer Science from the University of Minnesota, Minneapolis, USA with a thesis on 'Data Mining Techniques for Sequences and Graphs'. He talks on a range of issues flanking BI today.

When it comes to day-to-day business, what role does or can BI play? How much information derived in BI is indeed actionable?

BI is important because the company wants to know about the customer, wants to analyse the competition, wants to study customer churn properly for corrective action, specially in context to today's scenario and the crowded marketplace wherein competition has become more fierce.

Churn identification, for instance, helps at the decision-making level directly for customer or campaign management. It's done by churning out reports, dashboards, predictive analysis etc from data warehouses.

How best can BI be applied to broader issues like national security?

Security is a hard management problem. It's all about how best to identify and focus on the priority areas. It's all about expanding relationships. Information in security incidents is mostly divided and scattered, but it's there. You have to find how best to link them.

The inference part is where the human comes in. But linking and tracing information is where technology can play a role. Today terror attacks are planned and managed often on Internet. What can be done better is that in reflection, one can connect the dots and we can have a frame of reference.

If we take a frank look at BI's evolution so far, how successfully has the reactive to proactive gap been covered?

Yes, EPA (Enterprise Performance Management) is where we are doing proactive work. Unlike the earlier report-generating work and focus on time-cutting, there is need for proactive work. A lot of BI is still reactive but EPM is a new breed. BI is trying to look at decisions the other way round with stuff like 'what-if' questions and scenario-based analysis.

What's the connection status, between BI and higher realms like human intelligence, artificial intelligence, robotics, cybernetics etc?

These are advanced analytics areas whereas BI is an analytics area focusing specifically on data and business problems. Its focus is mainly on decision-making.

With new scenarios like clouds and grids redefining computing and data, how does BI realign itself?

BI may go to cloud or Grid if privacy and security are taken care of. That's a challenge. But for that big leap in mindset is required. More so on areas like support and control. Guys like Google and Amazon can afford the infrastructure to do it.

What's your take on future trends and the status of BI in India?

As we move ahead, BI would graduate to real-time BI. The quicker, the better. DWA (Data Warehousing Applications) is another area that helps in how to deal in hardware to solve big data problems.

Apart from that, open source BI would be something to watch out for. The way open source has made inroads in areas of OS, Internet browser etc with Linux or Firefox, we have to see what happens in BI space.

Talking of India, the problem here is that unlike in the West, there is no unique identifier for data like the SSN in the US. I heard it's about to come in India too, and if that happens, it will take away the identification problem for BI. Not all data is captured here. Industry is still tactical in its approach towards BI.

What has been Persistent's ground of action so far in BI?

It's been traditionally an area of database. We have focused on building tools. Our work happens around BI tool vendors, and product development for some major DB and DWH vendors, which is the OPD flavour. We also build lots of applications and tools. The spectrum so far covers product engineering for leading ETL vendors where we have data cleaning, de-duping.

We also cover predictive analysis consulting area with churn analysis, targeted advertising, risk-modeling and cross-selling as some constituents. Apart from this, we have various domain-specific practices in industries like telecom, retail etc for reporting, dashboards.