Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

GLM Tweedie dispersion parameter estimation only failed with dataset #15958

Closed
wendycwong opened this issue Nov 27, 2023 · 0 comments
Closed

GLM Tweedie dispersion parameter estimation only failed with dataset #15958

wendycwong opened this issue Nov 27, 2023 · 0 comments
Assignees
@tomasfryda
Labels
bug
Milestone
3.44.0.3

Comments

@wendycwong
Copy link
Contributor

wendycwong commented Nov 27, 2023 •
edited

Support ticket: https://support.h2o.ai/a/tickets/106503

For maximum likelihood estimation of dispersion parameter only for GLM tweedie family, the algo produced the wrong result for the dataset generated in the support ticket.

Regarding search algo, I will propose to try out Torczon’s multi-directional method. It avoids going into degenerate solutions.
@wendycwong wendycwong added the bug label Nov 27, 2023
@tomasfryda tomasfryda mentioned this issue Nov 30, 2023
Merged
@tomasfryda tomasfryda added this to the 3.44.0.3 milestone Dec 1, 2023
tomasfryda added a commit that referenced this issue Dec 15, 2023
@tomasfryda
GH-15958: GLM fix Tweedie ML dispersion estimation (#15959)
bfc1bd0 
* GLM fix Tweedie ML dispersion estimation

* Set dispersion estimated from var power and dispersion

* Fix case with no weight column and an R test so it uses as.h2o instead of as.h2o.data.frame

* Fix comment and disable tests for now
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@tomasfryda tomasfryda

Labels
bug
Projects
None yet
Milestone
3.44.0.3
Development

No branches or pull requests
3 participants
@mn-mikke @wendycwong @tomasfryda