[pymvpa] RFE question 2.0

[James M. Hughes]

Sorry to spam the list, but is is possible that there's a problem  
using RFE w/ CrossValidatedTransferError?

Here's some code I'm attemping to run:

rfesvm_split = LinearCSVMC()
     FeatureSelection =  
RFE(sensitivity_analyzer=rfesvm_split.getSensitivityAnalyzer(),  
transfer_error=TransferError(rfesvm_split), \
         feature_selector=FractionTailSelector(percent / 100.0,  
mode='select', tail='upper'), update_sensitivity=True)

     clf = FeatureSelectionClassifier(clf = LinearCSVMC(), \
         # on features selected via RFE
         feature_selection = FeatureSelection)
         # update sensitivity at each step (since we're not using the  
same CLF as sensitivity analyzer)

     clf.states.enable('feature_ids')

     cv = CrossValidatedTransferError(TransferError(clf),  
NFoldSplitter(cvtype=1), enable_states=['confusion'])

     error = cv(dataset)

     print 'Error:  ' + `error`

     return clf.confusion, clf.feature_ids


where dataset and percent are input arguments...

I get the following error (somewhere pretty far down the call stack):

    1060
-> 1061         predictions = self.clf.predict(testdataset.samples)
         predictions = undefined
         self.clf.predict = <bound method LinearCSVMC.predict of  
LinearCSVMC(kernel_type='linear', svm_impl='C_SVC')>
         testdataset.samples = undefined
    1062
    1063         # compute confusion matrix

<type 'exceptions.AttributeError'>: 'NoneType' object has no attribute  
'samples'


It seems like the testdataset argument, which is typically empty, in  
the FeatureSelectionClassifier is somehow not getting set correctly.   
I thought the splitter in CrossValidatedTransferError would take care  
of this.

Am I missing something big here?

Thanks for any help!

Cheers,
James.