News

Arnie Goodman's article Commentary on KDD-2001 or what is Data Mining and Statistics , (http://www.kdnuggets.com/news/2001/n24/1i.html) has provoked many responses.

Ok, I admit -- not all data miners know as much statistics as they should. Why? Plainly put, statistics are hard and many people (not just data miners) don't know them well. However, we can all agree that statistics are very useful and data miners should learn more statistics and avoid rediscovering statistics methods. Perhaps, part of the burden is on statisticians to better teach the most relevant theory to practitioners.

However, statistics is an important but small part of the entire process of knowledge discovery. Many other steps, including knowledge acquisition and data preprocessing are needed which are knowledge-based.

In many cases knowledge-based methods work better than statistical ones. For example, people don't burn their fingers 1000 times to learn not to put a finger in the fire.

So let's combine statistics, knowledge-based methods, and other approaches! Happy Discoveries!

See additional interesting and spirited responses in this issue.