Orange Forum • View topic - continuous features problems (MeasureAttribute_*)

continuous features problems (MeasureAttribute_*)

Report bugs (or imagined bugs).
(Archived/read-only, please use our ticketing system for reporting bugs and their discussion.)
Forum rules
Archived/read-only, please use our ticketing system for reporting bugs and their discussion.

continuous features problems (MeasureAttribute_*)

Postby emanuele » Tue May 02, 2006 15:56

Given this example:
Code: Select all
import orange, orngFSS
data = orange.ExampleTable("housing")
relief = orange.MeasureAttribute_relief()
mse = orange.MeasureAttribute_MSE()
info = orange.MeasureAttribute_info()
measure = relief
m = orngFSS.attMeasure(data,measure)
print m

If measure=relief I got strage results (i.e. all m values are negative or zero) with all dataset with continuous features. If measure=mse I got an exception telling that MSE is not good for continuous features (????). If measure=info I got a Python "Abort".

Since I need to use feature selection module with numerical and continuous values, what can I use?

Thanks

Emanuele

Postby emanuele » Tue May 02, 2006 15:58

More info:
orange-source-snap-2006-04-25.tgz
compiled on GNU/Linux intel 32bit.

Cheers,

Emanuele

Postby Blaz » Tue May 02, 2006 19:06

For MSE reply is right, since it should only work with continuous class and discrete attribute - you can estimate MSE for a set of instances that are cover with some value of a discrete features, you can not do this for a point in a continuous scale. For ReliefF and other measures we will look at your report in the next few days (it was a holiday time here since last Wed).


Return to Bugs



cron