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

Allow model built earlier to be used to calculate full loglikelihood and AIC for GLM #15809

Closed
wendycwong opened this issue Oct 9, 2023 · 0 comments · Fixed by #16025
Closed

Allow model built earlier to be used to calculate full loglikelihood and AIC for GLM #15809

wendycwong opened this issue Oct 9, 2023 · 0 comments · Fixed by #16025
Assignees
@syzonyuliia
Labels
feature
Milestone
3.46.0.1

Comments

@wendycwong
Copy link
Contributor

here is the support ticket: https://support.h2o.ai/a/tickets/106503

Basically, a customer would like to be able to calculate the full AIC and log likelihood function after a model has been built earlier perhaps using an earlier version of H2O-3.

Here is a proposal:

  1. Build a model with compute_p_values=True;
  2. Save the model to mojo (make sure dispersion parameter is saved to mojo and is available);
  3. Load the model back using generic_model and just call whatever function there is to calculate the full AIC and compute the full log likelihood.
@krasinski krasinski added this to the 3.46.0.1 milestone Oct 17, 2023
syzonyuliia pushed a commit that referenced this issue Jan 17, 2024
@syzonyuliia-h2o
GH-15809: implement AIC and Loglikelihood calculation for GenericModel
e0dc8e8
syzonyuliia pushed a commit that referenced this issue Jan 17, 2024
@syzonyuliia-h2o
GH-15809: add AIC and Loglikelihood to ModelMetricsBinomial
0ff937d
syzonyuliia pushed a commit that referenced this issue Jan 17, 2024
@syzonyuliia-h2o
GH-15809: implement AIC and Loglikelihood calculations for ModelMetri…
566eb00 
…csBinomial
syzonyuliia pushed a commit that referenced this issue Jan 17, 2024
@syzonyuliia-h2o
GH-15809: add AIC and Loglikelihood to output metrics
8f66d3e
syzonyuliia pushed a commit that referenced this issue Jan 17, 2024
@syzonyuliia-h2o
GH-15809: update test to check AIC and Loglikelihood calculation for …
8e86ba2 
…loaded model
syzonyuliia pushed a commit that referenced this issue Jan 22, 2024
@syzonyuliia-h2o
GH-15809: correct betas source
003d8f9
syzonyuliia pushed a commit that referenced this issue Jan 22, 2024
@syzonyuliia-h2o
GH-15809: implement AIC and loglikelihood calculation for multinomial…
b1d9b70 
… generic glm
syzonyuliia pushed a commit that referenced this issue Jan 22, 2024
@syzonyuliia-h2o
GH-15809: minor aic retrieval fix
7b4bebf
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: enable loglikelihood and AIC calculation for multinomial fa…
24fbf34 
…mily
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: remove prints
d5b7fc3
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: refactor
78a60a8
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: add new parameter to the constructor, and add new constructor
3a45e5c
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: add dispersion_estimated parameter to GLM mojo
49ba70a
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: update and fix tests
a084b0c
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: fix metrics exposure in python
4aafa03
syzonyuliia pushed a commit that referenced this issue Jan 26, 2024
@syzonyuliia-h2o
GH-15809: fix parameters
635b9fd
syzonyuliia pushed a commit that referenced this issue Jan 29, 2024
@syzonyuliia-h2o
GH-15809: add null check
cbad0f8
syzonyuliia pushed a commit that referenced this issue Jan 29, 2024
@syzonyuliia-h2o
GH-15809: fix tests
026f45d
syzonyuliia pushed a commit that referenced this issue Jan 31, 2024
@syzonyuliia-h2o
GH-15809: fix R tests
c909e14
@syzonyuliia syzonyuliia linked a pull request Jan 31, 2024 that will close this issue
GH-15809: fixes loglikelihood and aic for glm generic model #16025
Merged
syzonyuliia pushed a commit that referenced this issue Feb 1, 2024
@syzonyuliia-h2o
GH-15809: fix reading new parameter in MOJO load
4bbabef
syzonyuliia pushed a commit that referenced this issue Feb 1, 2024
@syzonyuliia-h2o
GH-15809: fix writing new parameter in MOJO load
de31362
syzonyuliia pushed a commit that referenced this issue Feb 1, 2024
@syzonyuliia-h2o
GH-15809: fix value
667b226
@syzonyuliia syzonyuliia mentioned this issue Feb 1, 2024
Merged
syzonyuliia pushed a commit that referenced this issue Feb 2, 2024
@syzonyuliia-h2o
GH-15809: implement AIC and Loglikelihood calculation for GenericModel
04e9c92
syzonyuliia pushed a commit that referenced this issue Feb 2, 2024
@syzonyuliia-h2o
GH-15809: add AIC and Loglikelihood to ModelMetricsBinomial
01e4d89
syzonyuliia pushed a commit that referenced this issue Feb 2, 2024
@syzonyuliia-h2o
GH-15809: implement AIC and Loglikelihood calculations for ModelMetri…
149ccd3 
…csBinomial
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: update test to check AIC and Loglikelihood calculation for …
d5dcae1 
…loaded model
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: correct betas source
7b4d93f
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: implement AIC and loglikelihood calculation for multinomial…
c20c2d5 
… generic glm
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: minor aic retrieval fix
455587d
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: enable loglikelihood and AIC calculation for multinomial fa…
50fb799 
…mily
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: remove prints
c97585a
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: refactor
749dd43
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: add new parameter to the constructor, and add new constructor
25ba898
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: add dispersion_estimated parameter to GLM mojo
db5463a
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: update and fix tests
a7269e4
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix metrics exposure in python
da44bfe
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix parameters
1dac03e
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: add null check
8665036
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix tests
eb0b133
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix R tests
be425a7
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix reading new parameter in MOJO load
ac8d8ab
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix writing new parameter in MOJO load
48939fd
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix value
76e62f6
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix comments
b2318c8
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix printing metrics
c8f0487
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: remove commented code
211ba61
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: assign NaN instead of 0 as placeholder value for Loglikelihood
c08e2f5
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: default dispersion estimation set to 1
efbbd6a
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: clean test
607f0fa
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fix aic check in test
6c46408
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: additionally fix aic check in test
22c383a
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: additionally fix aic check in test
a6c8bda
wendycwong pushed a commit that referenced this issue Feb 5, 2024
@syzonyuliia-h2o
GH-15809: fit test - add default parameters
4a90a0b
wendycwong pushed a commit that referenced this issue Feb 12, 2024
@syzonyuliia @syzonyuliia-h2o
GH-15809: fixes loglikelihood and aic for glm generic model (#16025)
a6bd451 
* GH-15809: implement AIC and Loglikelihood calculation for GenericModel

* GH-15809: add AIC and Loglikelihood to ModelMetricsBinomial

* GH-15809: implement AIC and Loglikelihood calculations for ModelMetricsBinomial

* GH-15809: add AIC and Loglikelihood to output metrics

* GH-15809: update test to check AIC and Loglikelihood calculation for loaded model

* GH-15809: correct betas source

* GH-15809: implement AIC and loglikelihood calculation for multinomial generic glm

* GH-15809: minor aic retrieval fix

* GH-15809: enable loglikelihood and AIC calculation for multinomial family

* GH-15809: remove prints

* GH-15809: refactor

* GH-15809: add new parameter to the constructor, and add new constructor

* GH-15809: add dispersion_estimated parameter to GLM mojo

* GH-15809: update and fix tests

* GH-15809: fix metrics exposure in python

* GH-15809: fix parameters

* GH-15809: add null check

* GH-15809: fix tests

* GH-15809: fix R tests

* GH-15809: fix reading new parameter in MOJO load

* GH-15809: fix writing new parameter in MOJO load

* GH-15809: fix value

* GH-15809: fix comments

* GH-15809: fix printing metrics

* GH-15809: remove commented code

* GH-15809: assign NaN instead of 0 as placeholder value for Loglikelihood

* GH-15809: default dispersion estimation set to 1

* GH-15809: clean test

* GH-15809: fix aic check in test

* GH-15809: additionally fix aic check in test

* GH-15809: additionally fix aic check in test

* GH-15809: fit test - add default parameters

* Fixed test discrepancies.

* only return AIC and loglikelihood for glm models

* fixed AIC problem when model is not glm

* Incorporate Tomas F review.

* replace m != null && m.getClass().toString().contains(generic) with score4Generic

---------

Co-authored-by: syzonyuliia <[email protected]>
Co-authored-by: wendycwong <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@syzonyuliia syzonyuliia

Labels
feature
Projects
None yet
Milestone
3.46.0.1
Development

Successfully merging a pull request may close this issue.

GH-15809: fixes loglikelihood and aic for glm generic model h2oai/h2o-3
3 participants
@wendycwong @krasinski @syzonyuliia