orvindemsy / bciciv2a-fbcsp Goto Github PK

I went through the code as I was interested in the FBCSP. But later at the end of the coding, it seems that under "Results for test data",
you have
data_train = mod_data[subj]['train']['mutual']
X_train = data_train['X']
y_train = data_train['y']

data_test = mod_data[subj]['test']['mutual']
X_test = data_train['X']     -------  X_test = data_test['X']
y_test = data_train['y']      -------  y_test = data_test['y']

From the training data results, it seems that the model might have been overfitted.

I did extend your work to 9 subjects using SVC, Naive Bayers Classifier and DecisionTrees.

I only get .gdf data

Create new key 'EEG_filtered' to store filtered EEG of each subject seem to be missing something.
for i in range(1, ns):
mod_data
Can you complete this code?

There appears to be a mistake in the test data evaluation figure in the README.
In cell 72 of the notebook, training data is used to obtain test scores.

X_test = data_train['X']
y_test = data_train['y']

Best regards!