vets_modelspec: Model Specification

In tsmodels/tsvets: Vector ETS Models

Model Specification

Description

Specifies an vector ETS model prior to estimation.

Usage

vets_modelspec(
  y,
  level = c("constant", "diagonal", "common", "full", "grouped"),
  slope = c("none", "constant", "common", "diagonal", "full", "grouped"),
  damped = c("none", "common", "diagonal", "full", "grouped"),
  seasonal = c("none", "common", "diagonal", "full", "grouped"),
  group = NULL,
  xreg = NULL,
  xreg_include = NULL,
  frequency = 1,
  transformation = "box-cox",
  lambda = NULL,
  lower = 0,
  upper = 1,
  dependence = c("diagonal", "full", "equicorrelation", "shrinkage")
)

Arguments

y	an xts matrix.
level	dynamics for the level component.
slope	dynamics for the slope component.
damped	dynamics for the dampening component.
seasonal	dynamics for the seasonal component.
group	a vector of indices denoting which group the series belongs to (when using the grouped dynamics).
xreg	an xts matrix of external regressors.
xreg_include	a matrix of dimension ncol(y) by ncol(xreg) populated with either 0, 1 or 2+ (0 = no beta, 1 = individual beta and 2 = grouped beta). It is also possible to have group wise pooling. For instance 2 variables sharing one pooled estimates, and 3 other variables sharing another grouped estimate would have values of (2,2,3,3,3). The index for group wise pooling starts at 2 and should be incremented for each new group added.
frequency	seasonal frequency of the series.
transformation	a valid transformation for y from the “tstransform” function in the “tsaux” package (currently box-cox or logit are available).
lambda	the Box Cox power transformation vector (see box_cox) in the tsaux package. If a single NA, then it will calculate optimal lambda based on the multivariate Box Cox approach of Velila (1993), else if a vector of NA values it will calculate the individual Box Cox optimal parameters. Can also be either a single value (common lambda) or vector of values (individual lambda).
lower	lower bound for the transformation.
upper	upper bound for the transformation.
dependence	dependence structure to impose.

Details

The specification allows to specify a vector additive damped ETS model with options for the dynamics of the states and dependence.

Value

An object of class “tsvets.spec” with the following slots:

target	A list with original data series, the data series index and the sampling frequency
transform	A list with details on the transformation
model	A list with details the type of model dynamics
dependence	A list with details about the dependence structure
xreg	A list with details on the external regressors
vets_env	An environment with pre-calculated state matrices and other parameters which will be passed to the estimation routine

References

Athanasopoulos, G and de Silva, A. (2012),Multivariate Exponential Smoothing for Forecasting Tourist Arrivals, Journal of Travel Research 51(5) 640–-652.
de Silva, A., R. Hyndman, and R. D. Snyder. (2010).The Vector Innovations Structural Time Series Framework: A Simple Approach to Multivariate Forecasting, Statistical Modelling (10) 353–74.

tsmodels/tsvets documentation built on June 13, 2022, 2:14 p.m.