search

fabsig / GPBoost

Combining tree-boosting with Gaussian process and mixed effects models
Other
574 stars 46 forks source link

cannot save gp_model #5

Closed aprilffff closed 3 years ago

aprilffff commented 4 years ago

as discussed in https://github.com/fabsig/GPBoost/issues/4#issuecomment-643186989

fabsig commented 4 years ago

Thanks for mentioning this. I have added a warning when saving a Booster that the GPModel (gp_model) is not saved.

I cannot tell you when I have time to implement this feature. Any contributions are highly welcome.

Ferran-pv commented 3 years ago

I'm also having problems trying to save the model.

KevinYanesG commented 3 years ago

First of all: nice package! I was following this tutorial and I encountered the problem that the SHAP package only analyzes the LightGBM part of the model. To the best of my knowledge, SHAP does not take into account the random effects part. It has probably to do with this issue, too.

fabsig commented 3 years ago

First of all: nice package! I was following this tutorial and I encountered the problem that the SHAP package only analyzes the LightGBM part of the model. To the best of my knowledge, SHAP does not take into account the random effects part. It has probably to do with this issue, too.

Thank you. Yes, only the tree-part of the model is interpreted and random effects are not taken into account. This is how it should be and this is not related to this issue of saving a model. There is not much to explain concerning the random effects. If you like, you can inspect every random effect of every group, but there are no interactions between the random effects and the other predictor variables.

Saving a model is currently not implemented. Implementing this requires additional code that does the saving of everything related to the gp_model (for the tree-part, one can use the LightGBM code).

This feature is on my todo list, but there are other things that have higher priority.

fabsig commented 3 years ago

As of version 0.4.0, it is now possible to save models in the R and Python packages. See these Python and R examples on how to do this.

Ferran-pv commented 3 years ago

Thanks!

On Thu, Feb 18, 2021 at 9:42 AM Fabio Sigrist notifications@github.com wrote:

Closed #5 https://github.com/fabsig/GPBoost/issues/5.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/fabsig/GPBoost/issues/5#event-4345489914, or unsubscribe https://github.com/notifications/unsubscribe-auth/AG4HV7UB5HMERUFZF4U63O3S7THFXANCNFSM4N4GBIIA .