-
-
Notifications
You must be signed in to change notification settings - Fork 40
Open
Labels
Docs 📚Something to be adressed in docs and/or vignettesSomething to be adressed in docs and/or vignettes
Description
To add support for models, especially for parameters and including all features, we need following methods:
insight::find_formula()
insight::model_info()
insight::get_parameters()
insight::get_statistic()
insight::find_statistic()
insight::get_varcov()
insight::get_df()
In other packages, these are optional, when the above methods work fine:
parameters::ci()
parameters::p_value()
parameters::standard_error()
orinsight::get_varcov()
Nice to have (though some could already work if the above methods are implemented):
insight::find_parameters()
insight::get_modelmatrix()
insight::get_predicted()
insight::link_inverse()
andinsight::link_function()
insight::get_residuals()
The simple way is just adding a method for parameters::model_parameters()
, however, than this model-class is not supported by all functions we offer in the easystats eco-system.
I think we need a vignette on the API, explaining how to add methods/support for new models, similar to what @mattansb has done here: https://easystats.github.io/effectsize/articles/effectsize_API.html
Maybe we can split those vignettes into two (or more?) parts:
- What to do to add new methods for extracting/tidying parameters? That would only require a method for
model_parameters()
. - What to do to add new methods for some easystats support? (I'm thinking of
check_model()
, which seems to be often used by users) This would probably require some more methods - must check. - What to do to add new methods for full easystats support? This would need methods in insight, maybe also parameters or bayestestR.
Metadata
Metadata
Assignees
Labels
Docs 📚Something to be adressed in docs and/or vignettesSomething to be adressed in docs and/or vignettes