By processing vast amounts of data, computers helped astronomers make new discoveries about the universe. Now they're helping banks and other companies learn more about their customers.

As telescopes scan the heavens they generate huge amounts of data. Take the Hubble Space Telescope, for example. It produces about 1,000 gigabytes of data each year - enough to fill more than 200 million pages. Newer telescopes generate even more.

So what does all this data tell us? That's the question PPARC asked researchers at the University of Edinburgh more than 10 years ago.

All and nothing.

The answer is both all and nothing.

Like other scientific instruments, telescopes only give us the basic facts. They report everything they find - the positions of the objects in space, their brightness and more. And they report it all as accurately as they are able.

Unfortunately telescopes don't explain anything. Why are the stars where they are? What makes them move? Which stars are in which galaxy? And why does their brightness change? Telescopes haven't a clue.

To answer questions like these, astronomers have to put the facts together. Patterns in the data give them vital clues about how the universe works. The problem is that there is so much data to look through.

Distant galaxies.

Revealing insights.

To solve the problem, researchers at the University of Edinburgh put some of the world's most powerful computers to work. They needed them to run the most advanced data mining tools - computer programs expert at finding patterns in vast quantities of data.

The insights they generated helped astronomers enormously - so much so that data mining has become essential in astronomical research. Over the years, data mining tools have helped identify many objects in space that astronomers hadn't noticed before - new black holes and quasars, for example.

But even 10 years ago, it was clear data mining tools had many other uses. The researchers in Edinburgh soon found they could help companies like BT, Barclays Bank and BP learn new things about their businesses. They could even predict how their customers would behave.

By 1995 data mining had become a big business for the University - one mature enough to stand on its own. Quadstone Ltd was created to develop it still further.

Over the years, Quadstone took up more ideas created through the University's work for PPARC.

By 2005 it had become a world-leading supplier of tools that reveal and predict customer behaviour. It employed 38 staff in offices in Boston USA, London and Edinburgh and was generating revenues of almost £3 million a year.

In December 2005 the company was taken over by another British company, Portrait Software. The deal valued Quadstone at £3.15 million.

Quadstone's products are now sold under the Portrait Quadstone brand.

Predicting and protecting customers.

While you may not be familiar with data mining, it knows a lot about you. All sorts of organisations use it to learn more about people's habits and routines.

By looking carefully at data about past purchases, data mining tools can find interesting patterns. That customers who bought A and B usually went on to buy C, for example. Companies use such information to focus their advertising.

The tools are also used to detect credit card fraud. Card companies have lots of data about previous cases. This information can be 'mined' to reveal tell-tale signs - patterns that suggest a fraud is about to take place. By looking out for these signs in the future, the companies can reduce their losses.

Customer checkout.


Posted by: Sean    Source

---