statsmodels.tsa.statespace.kalman_filter.PredictionResults

class statsmodels.tsa.statespace.kalman_filter.PredictionResults(results, start, end, nstatic, ndynamic, nforecast)[source]

Results of in-sample and out-of-sample prediction for state space models generally

Parameters:
  • results (FilterResults) – Output from filtering, corresponding to the prediction desired
  • start (int) – Zero-indexed observation number at which to start forecasting, i.e., the first forecast will be at start.
  • end (int) – Zero-indexed observation number at which to end forecasting, i.e., the last forecast will be at end.
  • nstatic (int) – Number of in-sample static predictions (these are always the first elements of the prediction output).
  • ndynamic (int) – Number of in-sample dynamic predictions (these always follow the static predictions directly, and are directly followed by the forecasts).
  • nforecast (int) – Number of in-sample forecasts (these always follow the dynamic predictions directly).
npredictions

Number of observations in the predicted series; this is not necessarily the same as the number of observations in the original model from which prediction was performed.

Type:int
start

Zero-indexed observation number at which to start prediction, i.e., the first predict will be at start; this is relative to the original model from which prediction was performed.

Type:int
end

Zero-indexed observation number at which to end prediction, i.e., the last predict will be at end; this is relative to the original model from which prediction was performed.

Type:int
nstatic

Number of in-sample static predictions.

Type:int
ndynamic

Number of in-sample dynamic predictions.

Type:int
nforecast

Number of in-sample forecasts.

Type:int
endog

The observation vector.

Type:array
design

The design matrix, \(Z\).

Type:array
obs_intercept

The intercept for the observation equation, \(d\).

Type:array
obs_cov

The covariance matrix for the observation equation \(H\).

Type:array
transition

The transition matrix, \(T\).

Type:array
state_intercept

The intercept for the transition equation, \(c\).

Type:array
selection

The selection matrix, \(R\).

Type:array
state_cov

The covariance matrix for the state equation \(Q\).

Type:array
filtered_state

The filtered state vector at each time period.

Type:array
filtered_state_cov

The filtered state covariance matrix at each time period.

Type:array
predicted_state

The predicted state vector at each time period.

Type:array
predicted_state_cov

The predicted state covariance matrix at each time period.

Type:array
forecasts

The one-step-ahead forecasts of observations at each time period.

Type:array
forecasts_error

The forecast errors at each time period.

Type:array
forecasts_error_cov

The forecast error covariance matrices at each time period.

Type:array

Notes

The provided ranges must be conformable, meaning that it must be that end - start == nstatic + ndynamic + nforecast.

This class is essentially a view to the FilterResults object, but returning the appropriate ranges for everything.

Methods

predict([start, end, dynamic]) In-sample and out-of-sample prediction for state space models generally
update_filter(kalman_filter) Update the filter results
update_representation(model[, only_options]) Update the results to match a given model

Attributes

filter_attributes
kalman_gain Kalman gain matrices
representation_attributes
standardized_forecasts_error Standardized forecast errors