"There are two sides to data mining, descriptive and predictive," says Dr. Fayyad. "Descriptive data mining reorganizes the data, digging deeper into it and pulling out patterns, such as customer similarity, which allows you to create a short description about that group of customers.
"Predictive data mining looks for the best prediction, such as the best product to pitch to a customer. You won't get much insight, but it increases the performance, the ROI. Using both techniques will give you the best results.
"An important issue today is SQL, the standard interface for databases, which has proven to be the wrong interface," Fayyad says. "As an example, let's say you worked for a telecommunications company, and you want to find records about cell phone fraud. Well, guess what? These naturally asked questions cannot be answered by today's databases, because the interface was designed to address problems where you know the target and you want the database to quickly retrieve the result. If you don't have an exact description of the target, you're lost with a database today. This is why data mining is seeing a lot of demand. 数据挖掘实验室
"When I started in this field back in 1989, there were many people in large corporations struggling with large data sets. And even though there's a lot of data out there, it's not necessarily the right kind. Also, there's big difference in the ability to store data and the ability to access it in a useful way.
