decomp: Unobserved components decomposition

In tfarima: Transfer Function and ARIMA Models

Unobserved components decomposition

Description

Estimates the unobserved components of a time series (trend, seasonal, cycle, stationary, and irregular) based on the structure of an underlying model. The estimation can be carried out through the UCARIMA representation, the state-space model (SSM) form, or via forward/backward forecasts.

Usage

## S3 method for class 'ssm'
decomp(mdl, tol = 1e-05, ...)

## S3 method for class 'tfm'
decomp(
  mdl,
  y = NULL,
  method = c("mixed", "forecast", "backcast"),
  envir = NULL,
  ...
)

## S3 method for class 'ucarima'
decomp(mdl, ...)

decomp(mdl, ...)

## S3 method for class 'um'
decomp(
  mdl,
  z = NULL,
  method = c("ucarima", "ucarima0", "ssm", "ssm0", "mixed", "forecast", "backcast"),
  envir = parent.frame(),
  ...
)

## S3 method for class 'um'
decomp(
  mdl,
  z = NULL,
  method = c("ucarima", "ucarima0", "ssm", "ssm0", "mixed", "forecast", "backcast"),
  envir = parent.frame(),
  ...
)

Arguments

mdl	An object of class um or tfm, representing a univariate ARIMA or transfer function model.
tol	numeric tolerance for classifying eigenvalues.
...	Additional arguments passed to internal methods.
y	an object of class ts.
method	Character string specifying the decomposition method. Options are: "ucarima", "ucarima0" – using the UCARIMA representation, without or with the canonical requirement. "ssm", "ssm0" – using the state-space model representation, with multiple sources of error (MSOE) or a single source of error (SSOE), respectively. "mixed" – combining forward and backward forecasts. "forecast", "backcast" – using the forward or backward eventual forecast function. The last three options are deprecated and will be removed in a future release.
envir	environment in which the function arguments are evaluated. If NULL the calling environment of this function will be used.
z	an object of class ts.

Details

The function applies the corresponding internal routines to estimate the components depending on the chosen method. For UCARIMA-based methods, the Wiener–Kolmogorov filter is used. For state-space approaches, a Kalman smoother is applied.

Value

A data.frame with the estimated unobserved components.

Examples

Z <- AirPassengers
um1 <- um(Z, i = list(1, c(1, 12)), ma = list(1, c(1, 12)), bc = TRUE)
uc1 <- decomp(um1, method = "ucarima")

tfarima documentation built on Nov. 5, 2025, 7:43 p.m.