StateSpaceTimeSeries.predict#

StateSpaceTimeSeries.predict(X=None, coords=None, out_of_sample=False)[source]#

Predict data given input X.

Parameters:

  • X (DataArray | None) – Input features with dims [“obs_ind”, “coeffs”]. Must have datetime coordinates on obs_ind for out-of-sample predictions, and must contain the covariate columns used at fit time when the model was fit with exogenous regressors. Not required for in-sample predictions.

  • coords (dict[str, Any] | None) – Not used directly, datetime extracted from X coordinates.

  • out_of_sample (bool | None) – If True, forecast future values. If False, return in-sample predictions.

Returns:

Posterior predictive samples with y_hat and mu.

Return type:

xr.DataTree