Regression Steps

regression_step(
    name: str,
    *,
    kind: Literal[
        "ols",
        "ridge",
        "ridge_gs",
        "lasso",
        "lasso_gs",
        "elastic_net",
        "elastic_net_gs",
    ] = "ols",
    y_source: InpSources,
    X_source: InpSources,
    filter_key: str = "filter",
    y_payload_key: str | None = None,
    x_payload_key: str | None = None,
    y_column: Sequence[int] | int | None = None,
    X_columns: Sequence[int] | slice | None = None,
    intercept: bool = True,
    burn_in: int = 0,
    drop_initial: bool = False,
    variables: Sequence[str] | None = None,
    **kind_kwargs: Any,
) -> MCStep

regression_step conducts a regression of y_source on X_source and stores the regression summary in MCPipelineResult.regression_summaries under the key name. The regression is conducted separately for each replication, and summary statistics are stored as traces across replications.

Kind Dispatch:

kind	Result Type	Required kind_kwargs
"ols"	OLSResult	none
"ridge"	RidgeResult	alpha
"ridge_gs"	RidgeResult	start, stop, num
"lasso"	LassoResult	alpha
"lasso_gs"	LassoResult	start, stop, num
"elastic_net"	ElasticNetResult	alpha, l1_ratio
"elastic_net_gs"	ElasticNetResult	start, stop, num, l1_ratio

Target must be 1D

regression_step does not support multivariate targets. Multiple target values are to be regressed on the same set of regressors, separate regression_step components should be appended to the pipeline for each target variable.

Forwarded Regression Arguments

kind_kwargs are passed directly to the selected regression function. Grid-search kinds also accept the underlying regression options such as criterion, max_iter, and tol when supported by that method.

Sources:

Source	Description
states	Use context.data.states. Set drop_initial=True to remove the initial state row.
observables	Use context.data.observables.
x_pred, x_filt, y_pred, y_filt, innov, std_innov	Read arrays from the FilterResult stored under filter_key.
payload	Read an array-like object from context.payloads[payload_key].