search

Xtra-Computing / thundergbm

ThunderGBM: Fast GBDTs and Random Forests on GPUs
Apache License 2.0
689 stars 84 forks source link

Can the Negative Weights be Input? #82

Open BoanSy opened 11 months ago

BoanSy commented 11 months ago

Dear authors,

can we input the negative weights into our original and target data simultaneously in thundergbm?

With best respect,

BoanSy

zeyiwen commented 10 months ago

BoanSy, negative target values are totally fine. Negative feature values are also acceptable.

BoanSy commented 10 months ago

Thanks for your reply, but I'd like to ask how can we input them? Like clf.fit(original, target, sample_weight=weights)?