Beware, clairvoyants and fortune-tellers! Meet a young man named Timo Sulg (pictured) who predicts the future using algorithms.

Timo Sulg is usually, like most IT people, anything but flamboyant. He studies at Tallinn University of Technology and will graduate in two years. What is unusual about him is the fact that whereas students at his institution usually begin programming or go into the Web business, Timo is a miner. A data miner!

Some time ago, Timo discovered that modern branch of computer science, which is not that closely dealt with in Estonia. Since that time, he has absorbed a great amount of information, and read a great number of scholarly articles. He submitted his business idea to the StartITup! competition organized by Tehnopol and Tartu Science Park and now is polishing it with a number of other people!

But what’s it all about?

Data mining is analysis of large set of data using various algorithms and models to find concrete trends or patterns. There is a huge amount of data in the world: what we buy, where we go, what we click on, etc. For instance, if you go through all of the purchases made in Selver stores in the course of one month, you can see what people prefer on certain days and the product selection can be designed accordingly. Banks can easily offer new loan products if they use data mining. Insurance companies can use it to analyze actuarial data.

By analyzing our purchases, we can see the ways in which we can save money. “I want to create a practical analytical platform that can be used not only by big corporations but also small companies and individuals so that useful information could be extracted from a large set of data,” says Sulg.

Timo is not starting from scratch. He currently works for the Finnish company HRM, analyzing the movements and motivation of the employees. “I look for patterns. I get the algorithms ready and let them go to work on the data, while drinking coffee,” he says.

Algorithms already know the answers

He has also viewed changes in the price of crude oil on the world market, but says he hasn’t yet uncovered anything surprising about the future. It is probably no news to anyone that the price of oil has nowhere to go but up. But what could the optimum electricity production be for a country in order to achieve the best possible price level? The algorithms in Timo’s computer know.

He is in principle ready to predict the results of the spring 2011 elections in Estonia as well. It is a separate question as to what data to feed the algorithms – last election’s results, level of political activity in parliament or something else? The more variables, the more reliable the patterns that emerge will be. It might happen that data from the cheese-making sector in Bangladesh will correctly predict gainers and losers on NASDAQ, but this, says Timo, can be put down to blind chance.

Timo says that his clients in the future need not be people but other computers, as the analysis of the data could take place automatically.

Raido Pikkar, who advises Timo at Tehnopol, sees the young man’s project as a small step toward self-evolving artificial intelligence, something former Skype engineer Jaan Tallinn has also written about in the Estonian media recently.

Photo: Tiit Blaat/Eesti Ekspress