<div dir="ltr">Dear all,<div><br></div><div>I'm confused on how to pass selected features from SVM sensitivity analysis into another SVM for cross validation and performance measurement.I use HalfPartitioner to tell the SVM to divide data into training and testing groups based on 'chunks'. Here is my script</div>

<div><br></div><div># from mvpa2.tutorial_suite import *</div><div># clf=LinearCSVMC()</div><div># attr = SampleAttributes(os.path.join('/my/path/', 'attributes.txt'))</div><div>
# ds= fmri_dataset(samples='/path/to/samples.nii.gz', targets=attr.targets, chunks=attr.chunks )</div><div># fsel = SensitivityBasedFeatureSelection(clf, FixedNElementTailSelector(1000, mode='select', tail='upper'))</div>

<div># fclf = FeatureSelectionClassifier(clf, fsel)</div><div># cvte = CrossValidation(fclf, HalfPartitioner(attr='chunks'), enable_ca=['stats'])</div><div># results = cvte(ds)</div><div>
 </div><div>It ends up with this error:</div><div><br></div><div><pre style="vertical-align:baseline;outline-style:none;font-size:inherit;font-style:inherit;font-variant:inherit;font-weight:inherit;line-height:inherit;white-space:pre-wrap;margin-bottom:0px;font-family:monospace,sans-serif;margin-top:0px;border:0px;padding:0px">
<span style="margin:0px;padding:0px;border:0px;font-family:inherit;font-size:inherit;font-style:inherit;font-variant:inherit;font-weight:inherit;line-height:inherit;vertical-align:baseline;color:rgb(139,0,0)">ValueError</span>: Feature selectors cannot handle multiple sequences in a Dataset at once.  We got dataset <Dataset: 35x1@|S1, <sa: targets>> as input</pre>

<pre style="vertical-align:baseline;outline-style:none;font-size:inherit;font-style:inherit;font-variant:inherit;font-weight:inherit;line-height:inherit;white-space:pre-wrap;margin-bottom:0px;font-family:monospace,sans-serif;margin-top:0px;border:0px;padding:0px">
<br></pre>
<pre style="vertical-align:baseline;outline-style:none;font-size:inherit;font-style:inherit;font-variant:inherit;font-weight:inherit;line-height:inherit;white-space:pre-wrap;margin-bottom:0px;font-family:monospace,sans-serif;margin-top:0px;border:0px;padding:0px">
I can work with OneWayAnova() feature selector (as described in the tutorial page) with no problem, but here I get confused when I try to generalize the idea to another classifier</pre><pre style="vertical-align:baseline;outline-style:none;font-size:inherit;font-style:inherit;font-variant:inherit;font-weight:inherit;line-height:inherit;white-space:pre-wrap;margin-bottom:0px;font-family:monospace,sans-serif;margin-top:0px;border:0px;padding:0px">
<br></pre><pre style="vertical-align:baseline;outline-style:none;font-size:inherit;font-style:inherit;font-variant:inherit;font-weight:inherit;line-height:inherit;white-space:pre-wrap;margin-bottom:0px;font-family:monospace,sans-serif;margin-top:0px;border:0px;padding:0px">
There are two useful discussion on the list regarding a similar problem here:</pre><pre style="vertical-align:baseline;line-height:inherit;outline-style:none;font-family:monospace,sans-serif;font-size:inherit;font-style:inherit;font-variant:inherit;font-weight:inherit;white-space:pre-wrap;margin-bottom:0px;margin-top:0px;border:0px;padding:0px">
<div style="font-family:arial;white-space:normal"><a href="http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/2008q3/000085.html" target="_blank">http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/2008q3/000085.html</a><br>
</div><div style="font-family:arial;white-space:normal"><br></div><div style="font-family:arial;white-space:normal">and here:</div><div style="font-family:arial;white-space:normal"><a href="http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/2012q3/002181.html" target="_blank">http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/2012q3/002181.html</a></div>
<div style="font-family:arial;white-space:normal"><br></div><div style="font-family:arial;white-space:normal">So I think I have done my homework, but I can not still make it to work.</div><div style="font-family:arial;white-space:normal">
<br></div><div style="font-family:arial;white-space:normal">Thanks in advance for any inputs. </div><div style="font-family:arial;white-space:normal"><br></div><div style="font-family:arial;white-space:normal">All the best,</div>
<div style="font-family:arial;white-space:normal">Arman </div><div style="font-family:arial;white-space:normal"><br></div><div style="font-family:arial;white-space:normal">p.s I use fmri_dataset above but in fact my data is pre-processed (and normalized) structural GM segments, and I have used the concept of chunk to divide data into groups. </div>
<div style="font-family:arial;white-space:normal"><br></div></pre>
</div></div>