machineJS

machineJS copied to clipboard

basically, make the makePrediction function usable within node, not just by passing in a new file

ClimbsRocks

add in polynomial interactions between features

1

this should make our linear estimators more effective and frankly removes much of the burden of feature engineering from the engineer.

ClimbsRocks

better parameters for xgboost for gridsearch

1

ClimbsRocks

http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.BaggingClassifier.html This may not work with gridsearchcv, so we might have to build in a little extra logic gate around that.

ClimbsRocks

ClimbsRocks

that will make it compatible with windows machines as well.

ClimbsRocks

one of the parameters expects an int but is getting some number that is not an int. it's just a warning, but it's annoying. see if we can figure out...

ClimbsRocks

seems they're re-organizing searchCV and feature selecting

ClimbsRocks

right now we're only doing this at the very end for the ensembled predictions, but try doing this for the stage 0 predictions as well. the code's already written, it's...

ClimbsRocks

since we're now using ensembling to pick which algos to use, we want as many as possible. so consider training the same total number of estimators, but making each individual...

ClimbsRocks