search

ClimbsRocks / machineJS

[UNMAINTAINED] Automated machine learning- just give it a data file! Check out the production-ready version of this project at ClimbsRocks/auto_ml
https://github.com/ClimbsRocks/auto_ml
408 stars 64 forks source link

MultinomialNB fails when features have negative values #176

Open sunnmy opened 8 years ago

sunnmy commented 8 years ago

On my dataset machineJS failed because it was trying to use negative X values with multinomial Naive Bayes classifier.

Suggested solutions:

ClimbsRocks commented 8 years ago

Sweet, thanks for pointing that out! Out of curiosity, did the entire process break at that point, or did the rest of it continue to run with only MultinomialNB failing?

Your suggested solutions are solid. We're already doing feature normalization in data-formatter, so it shouldn't be too hard to expand that here. The other two look very straightforward as well. If you wanted to take a crack at any of those solutions, that'd be an easy PR to merge!

Thanks for reporting this! Keep the issues coming :)

muffin-head commented 3 years ago

multinomial don't take neg values, so while preprocessing numerical data best use is minmaxscaler()