This error appears only during submission and not in the AD Workspace.
cat_encoder = load_object('cat_encoder')
X_cat = pd.DataFrame(cat_encoder.transform(test[['intersection_pos_rel_centre']]).toarray())
X_test = test.drop(['row_id','intersection_pos_rel_centre'],axis=1)
X_test = X_test.join(X_cat)
X_test.head()
-----------------
ValueError: Input contains NaN
The error appears here:
X_cat = pd.DataFrame(cat_encoder.transform(test[['intersection_pos_rel_centre']]).toarray())
The thing is that there are NaNs but I have checked and it is already a category, and this is not a problem in the AD Workspace but only in submission enviroment.