When doing multiregression, i.e. multiple target values in catboost regression,

Question

When doing multiregression, i.e. multiple target values in catboost regression,

is there a difference between using one model per target value vs using one model for all targets with loss function set as multiRMSE ? Or will both methods give the same predictive power for each target value?

#english #programming

0

06.01.2022

5 ответов

62 просмотра

Thomas Wolf

Tay
How does one implement multiple regression in catb...

With setting the loss function to MultiRmse or MultiRmsewithmissingvalues

0

09.01.2022

Driaan Jansen Автор вопроса

Tay
How does one implement multiple regression in catb...

https://catboost.ai/en/docs/concepts/loss-functions-multiregression#MultiRMSE Yip, just set loss function = "MultiRMSE" when you define your model.

0

09.01.2022

Tay

Driaan Jansen
https://catboost.ai/en/docs/concepts/loss-function...

I just saw people on forums complaining on how this link barely explains anything, I only found an old kaggle notebook for sample catboost multi regression code. So a friendly advice for you guys would be adding a basic sample code here. I think this is a huge feature and would put you ahead of many algorithms if people knew this existed.

0

09.01.2022

Thomas Wolf

Tay
I just saw people on forums complaining on how thi...

IT especially leacks the mention that it also works with missing values in the target variables, which makes it pretty unique

0

09.01.2022

Tay · Accepted Answer

Tay

How does one implement multiple regression in catboost?

0

08.01.2022

169 похожих чатов

When doing multiregression, i.e. multiple target values in catboost regression,

5 ответов

Похожие вопросы