I have a fMRI dataset which I divide into a train and test set. I create a decoder using
myDecoder = nilearn.decoding.Decoder(...standardize=True)
Then I use nilearn.decoding.Decoder.fit() to fit to the training set. Finally, I use nilearn.decoding.Decoder.predict() to predict classification in the test set.
Prior to training, the decoder has applied a certain mean and SD to each voxel in the dataset because I set standardize=True.
My question is–does it apply the same mean and SD, i.e., a comparable standardization, to the test set prior to classifying it as well?
Apologies if this seems like an obvious question–it seems like it certainly should, but I couldn’t find in the documentation and wasn’t able to verify myself in the code that this in fact takes place.
I am trying to troubleshoot some results and want to verify that the standardization is occurring correctly.