Replies: 1 comment 2 replies
-
Hi, I'm not familiar with longitudinal data analysis, any suggested reading?
Yes, otherwise, the Hessian becomes a 3-dim cube for the dataset, assuming targets are strongly associated. XGBoost currently handles only vector and matrix gradients.
The off-diagonal of the Hessian matrix?
Could you please share the math (equations)? |
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello!
I am very new to XGBoost, so please excuse me if I ignore some obvious piece of knowledge.
I am currently working on a problem with a longitudinal setting where I am trying to predict multiple outputs.
In longitudinal data, off-diagonal terms are essential for reflecting intra-subject correlation. However, I noticed that we should use a diagonal approximation for hessian in the custom loss.
I want to use deviance loss for my custom loss using a covariance matrix for each subject. Is there a way to use full hessian in the multioutput regression in XGBoost?
Thank you for your answer.
Beta Was this translation helpful? Give feedback.
All reactions