dataset: DATA GENERATING FUNCTIONS: This needs to be documented very...

In b-brune/tvRRR: Time-varying reduced-rank regression using state-space models

Description

DATA GENERATING FUNCTIONS: This needs to be documented very carefully

Usage

dataset(type, p, d, q, t, X, forecast, model = "A", ...)

Arguments

type	= c("deterministic", "break", "VARbreak", "rw", "fixed"), see Details for a description of the different kinds
p	dimension of the target vector
d	latent dimension of the model
q	dimension of the predictors
t	length of the time series
X	optional, the predictors used to generate the time series. Not used if type = "VARbreak". If not handed over, it is drawn randomly from a q-variate normal distribution with mean zero and covariance identity
forecast	how many additional variables do we want to generate for forecasting experiments
model	which model do we want? Either "A" or "B"
...	additional variables specific to the different types, see Details.

Details

Optional parameters that may be handed over to the function are

• alpha_0 optional

• alpha_1 optional (only used if model = "A" and type = c("break", "VARbreak", "deterministic") )

• beta_0 optional

• beta_1 optional (only used if model = "B" and type = c("break", "VARbreak", "deterministic") )

• Omega error covariance, defaults to the identity matrix

• shift only used if type = c("break", "VARbreak", "deterministic")), guides how far the matrices are apart (see Details)

• Sigma column covariance if type = "rw"

• Delta row covariance if type = "rw", defaults to the identity

• breakpoint position of the structural break in the dataset for type = c("break", "VARbreak")

Value

A named list with the desired dataset, containing components

y	the target variable (t x p) matrix
X	the predictors (t x q) matrix
alpha	if model = "A", an array of dimensions (t, p, d), if model = "B", a (p, d) matrix
beta	if model = "B", an array of dimensions (t, q, d), if model = "A", a (q, d) matrix
Omega	the error covariance (p, p) dimensional
Sigma	if type = "rw" the column covariance of the states, (d, d) matrix

b-brune/tvRRR documentation built on Dec. 19, 2021, 6:37 a.m.