I am using sample of (categorical and numerical features) for creating DL model in DLS.How ever after creating model when i tried to predict on test data it giving error that some categorical field has different categories which was not present on training data categorical field.
So my question is there an option to use any encoding technique as per of data processing before training model.Like different categories can be handled by some encoding technique in python scikit learn we have used when created model through python scitkit learn…Ideally this should be automatically handles by DLS…as H2o Flow(ML platform with UI interface) automatically handles these things like passing any new field in test data,new category in test data etc…
I did not found any such option in DLS .Can you please suggest how we can do this?
No sorry DLS doesn’t handle this problem right now. You need to have same categories.