pymargins.BootstrapOnlyAdapter

class pymargins.BootstrapOnlyAdapter

Base for non-parametric / algorithmic models with no meaningful Σ̂.

Covers tree ensembles, kNN, neural networks, and other algorithmic estimators. Declares only bootstrap support; the inference engine routes all requests for these models through refit-and-recompute.

Subclasses must implement refit().

__init__()

Methods

__init__()
attach(session)	Attach this adapter to a Margins session.
bootstrap_state()	Replay state for a refitted adapter.
coefficients()	Raise: bootstrap-only models have no meaningful coefficient vector.
column_index_of_variable(name)	Return the design-matrix column index corresponding to a variable.
covariance([vcov_spec])	Raise: bootstrap-only models have no meaningful covariance.
design_matrix_from_df(df)	Build a design matrix from a concrete DataFrame of evaluation rows.
predict(beta, X[, offset])	Raise: bootstrap-only models do not provide parametric predict.
refit(resampled_data, *[, index])	Refit the model on resampled data, returning a new adapter.
variable_metadata()	Return per-variable metadata used by averaging and validation.

Attributes

gradient_backend_recommendation	Not used (no delta path); declared for completeness.
n_outcomes	Number of outcome classes for multi-outcome models, default 1.
outcome_labels	Outcome class labels for multi-outcome models, or None.
supported_inference_methods	Bootstrap-only adapters declare only bootstrap support.
supports_jax_autodiff	Bootstrap-only adapters have no parametric predict; no autodiff.
training_data	The training data used to fit the model.