Re: another attempt

[http://dummy.us.eu.org/robert (Robert)]

 > From: Leonid <http://www.gmail.com/~l>
 > Date: Tue, 3 May 2005 22:00:28 -0400
 >
 > I looked at it in gnuplot (of course it is not allowed since
 > everything is supposed to happen automatically). It looks like 4
 > sausages coming close together at one end. If you have a sec., say
 > "gnuplot" and then splot "test.txt" (I'm attaching test.txt). Any
 > ideas what classifies such things well?

So, it has multiple maxima/minima?  Those are usually difficult.  I think
PVClus would have done an OK job, as probably would OC1.  Probably CART or
C4.5 would be OK, too.  Linear regression probably will not work well.
Logistic regression may.  Don't know anything about neural nets or GAs.

Re: gnuplot.  Apparently, I don't have gnuplot installed.  I'll just trust
your word w.r.t. the graph.

...I just looked at OC1.  Its input format is tab-separated.  There is
just a simple program mktree which outputs the decision tree.  Apparently,
the ftp is down.

 > L