Orange Forum • View topic - Orange Canvas: Feature weightings

Orange Canvas: Feature weightings

A place to ask questions about methods in Orange and how they are used and other general support.

Orange Canvas: Feature weightings

Postby avico510 » Thu Feb 23, 2012 22:36

I have been playing around with orange canvas (not the python script) for a bit now, and I really like it. But I was wondering if it is possible to get the feature weithings for say a logistic regression. Perhaps I am not using the right nomenclature, but something like the regression values, ie, if I have three variables a,b,c (and a ' y ' dependant variable ) to somehow get an output that would give me

y = x1(a) + x2(b) + x3(c) ..

is there a way to get this equation from the canvas? I cant seem to find it.

If not, perhaps you could point me towards how I could do this via the scripting .

Thank you kindly

Re: Orange Canvas: Feature weightings

Postby Ales » Mon Feb 27, 2012 10:45

You can't do this in the canvas, unless you use the Python Script widget. Connect the Logistic Regression widget to the Python Script widget (connect the classifier signals) and then execute this script
Code: Select all
print in_classifier.beta
in the Python Script widget to get the list of coefficients. You can also check the full documentation for Logistic Regression learner/classifier: http://orange.biolab.si/doc/reference/Orange.classification.logreg

Re: Orange Canvas: Feature weightings

Postby avico510 » Wed Feb 29, 2012 4:08

thank you ales... that was easy. Now i have to figure out how to do it for an svm but I will read up on the documentation.. hopefully I can figure it out. http://orange.biolab.si/doc/reference/O ... ation.svm/

Thanks again


Return to Questions & Support



cron