Warning message when the distribution is set to multinomial

gbm-developers / gbm

Gradient boosted models (the old gbm package)

Other

51 stars 27 forks source link

Warning message when the distribution is set to multinomial #62

Closed Epimorpha1 closed 8 months ago

Epimorpha1 commented 2 years ago

Hi, can you please advise me what to do when running gbm with categorical response variables and when the distribution is set to multinomial. When assuming a multinomial distribution this message is displayed:

"distribution = "multinomial"' is ill-advised as it is currently broken. It exists only for backwards compatibility. Use at your own risk.

What specifically is "broken" and can the results from the gbm be trusted?

gregridgeway commented 8 months ago

We never felt confident that the multinomial code was doing the right thing. There were a number of odd results that we decided to add the message. Then development moved over to gbm3 and we never went back to focus on multinomial again