Added MLP with a few commented out variations
Adjusted scaling again - so sklearn.preprocessing.scale gives a 0 mean and 1 variance and I think this is what we want (instead of .normalize). However, we need to scale the dev data using the training datas mean and variance (I think).
Added MLP with a few commented out variations Adjusted scaling again - so sklearn.preprocessing.scale gives a 0 mean and 1 variance and I think this is what we want (instead of .normalize). However, we need to scale the dev data using the training datas mean and variance (I think).