This is a package for state filtering, smoothing, and parameter estimation in state space models.

Provides methods for a state space model such as filtering (Kalman filter), smoothing (Kalman smoother), forecasting, likelihood evaluation, and estimation of hyperparameters (Maximum Likelihood, Expectation-Maximization (EM), and Expectation-Conditional Maximization (ECM), w/ and w/o penalization).

Currently only supports filtering, smoothing, and estimation for linear Gaussian state space models.

The following filter methods are supported :univariate, :collapsed, :multivariate, and :woodbury, which correspond to the following filter types

• UnivariateFilter: Filter using the univariate treatment for a linear Gaussian state space model.
• MultivariateFilter: Standard multivariate filter for a linear Gaussian state space model.
• WoodburyFilter: Same as MultivariateFilter, but uses the Woodbury identity to compute the inverse.

The smoother type is

• Smoother: General state smoothing, which accepts both multivariate and univariate filters and handles arbitrary state autocovariance smoothing.

Estimation can be done using Maximum Likelihood estimation and Expectation-Maximimxation algorithm. Both approaches can handle penalization/regularization on the parameters of the model.