Thanks so much for the reply! I took a look at the unit tests at <a href="http://nullege.com/codes/show/src%40p%40y%40PyMVPA-HEAD%40mvpa%40tests%40test_rfe.py/20/mvpa.featsel.rfe.RFE/python">http://nullege.com/codes/show/src%40p%40y%40PyMVPA-HEAD%40mvpa%40tests%40test_rfe.py/20/mvpa.featsel.rfe.RFE/python</a> .<br>
and the line (352) with SplitClassifier has been commented out...does that mean for 0.6.x versions of PyMVPA, SplitClassifier is no longer needed?<br><br>Please let me know if I've got this wrong...<br>In the 0.6.x. version:<br>
1. RFE uses sensitivies from the classifier to select features<br>2. meta-classifier FeatureSelectionClassifier trains an untrained version of the classifier on the selected features<br>3. CrossValidation uses that trained classifier on data partitioned by NFoldPartitioner (I have 6 runs/chunks, it splits one chunk out each time?)<br>
Thus we don't need SplitClassifier anymore?<br><br>Also, is it correct that I put Splitter('chunks')? i.e. in line 6:<br>fs = \<br> RFE(rfesvm_split.get_sensitivity_analyzer(),<br> ProxyMeasure(<br>
rfesvm_split, <br> postproc=BinaryFxNode(mean_mismatch_error, 'targets')),<br> Splitter('chunks'),<br> fselector=FractionTailSelector(<br> percent / 100.0,<br>
mode='select', tail='upper'), <br> update_sensitivity=True)<br><br>Still trying my best to get everything right! Thanks again for your time. :) Any input will be very much appreciated!<br>
<br>Kimberly<br><br>-----Original Message-----<br>
From: Yaroslav Halchenko [mailto:<a href="mailto:debian@onerussian.com">debian@onerussian.com</a>]<br>
Sent: Tuesday, June 14, 2011 5:34 PM<br>
To: Development and support pf PyMVPA<br>
Subject: Re: [pymvpa] RFE confusion_state<br>
<br>
sorry... didn't have change to look in detail -- I guess that doc page<br>
of ours got outdated since it reflects how you would do RFE using 0.4.x<br>
version of PyMVPA. So you could do it that way if you used 0.4.7.<br>
Sorry about the confusion.<br>
<br>
For 0.6.x -- here is an example from our unittests which I hope would be of<br>
help help. It does pretty much the same ;-)<br>
<br>
def test_james_problem(self):<br>
percent = 80<br>
dataset = datasets['uni2small']<br>
rfesvm_split = LinearCSVMC()<br>
fs = \<br>
RFE(rfesvm_split.get_<div id=":8t">sensitivity_analyzer(),<br>
ProxyMeasure(rfesvm_split,<br>
postproc=BinaryFxNode(mean_mismatch_error,<br>
'targets')),<br>
Splitter('train'),<br>
fselector=FractionTailSelector(<br>
percent / 100.0,<br>
mode='select', tail='upper'), update_sensitivity=True)<br>
<br>
clf = FeatureSelectionClassifier(<br>
LinearCSVMC(),<br>
# on features selected via RFE<br>
fs)<br>
# update sensitivity at each step (since we're not using the<br>
# same CLF as sensitivity analyzer)<br>
<br>
cv = CrossValidation(clf, NFoldPartitioner(), postproc=mean_sample(),<br>
enable_ca=['confusion'])<br>
<br>
<br>
if you want to see the progress of RFE -- just enable debug variables<br>
RFEC and RFEC_ for more verbosity... eg. in the environment<br>
<br>
export MVPA_DEBUG=RFEC,RFEC_<br>
<br>
before running your python script<br>
<br>
On Tue, 14 Jun 2011, Kimberly Zhou wrote:<br>
<br>
> Hi all,<br>
> I'm still trying to get recursive feature selection to work with the<br>
> example from<br>
> [1]<a href="http://dev.pymvpa.org/featsel.html#recursive-feature-elimination" target="_blank">http://dev.pymvpa.org/featsel.html#recursive-feature-elimination</a>,<br>
> however I can't get past an error from the line with<br>
> confusion_state="confusion".<br>
> Here's the code:<br>
--<br>
=------------------------------------------------------------------=<br>
Keep in touch <a href="http://www.onerussian.com/" target="_blank">www.onerussian.com</a><br>
Yaroslav Halchenko <a href="http://www.ohloh.net/accounts/yarikoptic" target="_blank">www.ohloh.net/accounts/yarikoptic</a><br>
<br>
_______________________________________________<br>
Pkg-ExpPsy-PyMVPA mailing list<br>
<a href="mailto:Pkg-ExpPsy-PyMVPA@lists.alioth.debian.org">Pkg-ExpPsy-PyMVPA@lists.alioth.debian.org</a><br>
<a href="http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa" target="_blank">http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa</a></div>