WHERE IS DATA MINING HEADED?
by John K. Thompson, Senior Director of Marketing, Magnify

I've just returned from the Data Mining Summit, where some of the best and brightest names in the business have been weighing in on where data mining is headed.

And here is the clearest message from the Summit: desktop data mining is fading, if in fact it was ever a truly viable force in the first place.

While the price of entry is low, companies that rely on desktop data mining tools will not get the value they need, or even the value they think they are getting. In doing so, they may ultimately sour themselves on the true worth of data mining.

Let me explain. The true value in data mining is in the ability to search for patterns in huge databases. You already know the uses: customer retention and optimization, fraud detection and prevention and so on. But as the cost of storage and data warehousing continue to drop, and the ability to store billions of records continues to grow, companies need more powerful tools to search and analyze the data for key patterns.

Desktop tools will not do the trick.

They will give you only a snapshot of the information you need, and the dangers of sampling versus a complete database analysis been well debated and documented.

Enterprise data mining needs powerful, enterprise-level tools. The ability to quickly share predictive models across various parts of the enterprise, and even between companies, will not be achieved using desktop-centric desk mining.

The promise of data mining has been to discover previously unknown or misunderstood insights. Relying on desktop tools places the responsibility on analysts to pick the area of inquiry. That sharply limits the search for those new insights. And that's not data mining: it's ROLAP at best, and a shame at worst.