Thanks very much.<div>I followed what yaroslav said and I got the p-values for each lights. The code is below:</div><div><div>-----------</div><div>permutator = AttributePermutator('targets', count=N)</div><div>distr_est = MCNullDist(permutator, tail='right', enable_ca=['dist_samples'])</div>
<div> </div><div>sl = Searchlight(cv,</div><div> IndexQueryEngine(voxel_indices=Sphere(0),</div><div> event_offsetidx=Sphere(10)),</div><div> postproc=mean_sample(),</div>
<div> roi_ids=np.arange(nfeatures),</div><div> null_dist=distr_est,</div><div> enable_ca=['stats']</div><div> )</div></div><div>-----------</div><div>The solution is to add node 'null_dist' for the Searshlight.</div>
<div><br></div><div>Now I have another problem:</div><div>Is there a quick way to get the predicted values for each light?</div><div><div>The cv is like below:</div><div>---------</div><div>cv = CrossValidation(clfer, partitioner,</div>
<div> errorfx=lambda p, t: np.corrcoef(p, t)[0],</div><div> enable_ca=['stats'])</div></div><div>---------</div><div>The error function I use is to get the correlation between prediction and true.</div>
<div>How can I get the prediction for each searchlight while I can get for one light in one fold via cv.ca.stats.summaries[0].sets[0][1]?</div><div><br></div><div>Best.</div><div><br></div><div><br></div><div>Xiangzhen</div>