Orange Forum • View topic - problem using svm classifiers in python scripting

problem using svm classifiers in python scripting

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.

problem using svm classifiers in python scripting

Postby jonna » Fri Dec 01, 2006 10:41

Dear Orange people,

We have downloaded and installed Orange 0.9.64 on our Red Hat system.
We are now trying to implement support for persistent SVM classifiers. This was achieved for the svms created through the orngSVM_Jakulin module. However, when starting working on the svms created through the orngSVM module, this module will not even return svm classifiers to the python layer. Actually, the python execution is stopped right after trying to create the classifier. Hence, in the following code;

data = orange.ExampleTable("iris.tab")
svmLearner = orngSVM.SVMLearner()
print svmLearner
svmClassifier = svmLearner(data)
print svmClassifier

nothing happens after calling svmLearner with data (<SVMClassifier instance> is not printed). The funny thing is that it appears to work in the Canvas. At least I get results from the TestLearners and Predictions widgets. Also, I have a windows installation (0.9.62) where the script above works.

Has anyone had similar problems with 0.9.64 under Red Hat?
Jonna

Problem solved

Postby jonna » Fri Dec 01, 2006 11:29

My problem described above is solved by compiling my 0.9.64 version with the svm.cpp file from 0.9.62. This makes the svm classifiers available in the python interface and svm appreas to still work in the Canvas.
:D Jonna


Return to Bugs



cron