In article <9lg89e$a0g$1@perki.connect.com.au>, Wise <weiz@spam-block.sympac.com.au> wrote:> Control systems that learn their job on the job are the promise of > fuzzy learning (or Fuzzy data mining) techniques. > > ... > > When you devise a system that creates rules from an analysis of > "raw" data that is part of the workings of a system you find that > the proliferation of contradictory rules is common place. > > An analysis of the Irises database, for example, by many Data Mining > methods highlights that classification of the plant by analysing > basic measurements of its component parts is fraught with > contradiction.
I've looked at the Iris data, as have many generations of statisticians. The main problem with this data is that it is so boring. Nothing interesting is going on. It's easily handled by any number of simple statistical techniques. If your data mining system produces results "fraught with contradiction", you need a better data mining system, not fuzzy logic. Radford Neal ---------------------------------------------------------------------------- Radford M. Neal radford@cs.utoronto.ca Dept. of Statistics and Dept. of Computer Science radford@utstat.utoronto.ca University of Toronto http://www.cs.utoronto.ca/~radford ----------------------------------------------------------------------------