Did any use deep learning sequence models in the hackathon ? Just curious at the performance. I had an LSTM model go to 786 in the private LB.
I only used ensembles. DL on my end hasn't worked so well and all the ensembles I've used have been pretty decent.
But DL is totally worth looking into
thanks for sharing your thoughts. For me the ensembles did not work well :(
I only used ensembles. DL on my end hasn't worked so well and all the ensembles I've used have been pretty decent.
But DL is totally worth looking into
thanks for sharing your thoughts. For me the ensembles did not work well :(