<p>Hi all,</p><p>I performed feature selection using anova and wanted the features back to NIfTI. I used the following comment lines but encountered an error:</p><p><br></p><p>clf = LinearCSVMC()</p><p>fsel = SensitivityBasedFeatureSelection(OneWayAnova(),</p><p>                                        FixedNElementTailSelector(2000, mode='select', tail='upper'))</p><p><br></p><p>fclf = FeatureSelectionClassifier(clf, fsel)</p><p><br></p><p>cvte = CrossValidation(fclf, NFoldPartitioner(),</p><p>                       errorfx=lambda p, t: np.mean(p == t),</p><p>                       enable_ca=['stats'])</p><p>cv_results = cvte(ds)</p><p><br></p><p>final_dataset = fclf.mapper.forward(ds)</p><p><br></p><p>print final_dataset</p><p><Dataset: 128x2000@float32, <sa: chunks,targets,time_coords,time_indices>, <fa: voxel_indices>, <a: imgaffine,imghdr,imgtype,mapper,voxel_dim,voxel_eldim>></p><p><br></p><p>nimg = map2nifti(final_dataset)</p><p>nimg.to_filename('features.nii')</p><p><font color="#cc0000">ValueError: array is not broadcastable to correct shape</font></p><p><br></p><p>Has anyone else had the same errors before?</p><p>Any advice will be greatly appreciated. Thanks.</p><p>Best,</p><p>Bin Liu</p>