simulation: Simulate From an INAR(p) Model

In rdmatheus/tsinteger: Autoregressive Models for Analysis of Integer-Valued Time Series

Description

Simulate from an INAR(p) model.

Usage

inar.sim(
  n,
  alpha,
  par,
  thinning = c("binomial", "nbinomial", "poisson"),
  innovation = "PO",
  n.start = NA
)

Arguments

n	A strictly positive integer given the length of the output series.
alpha	A vector of INAR coefficients.
par	A vector with the innovation process parameters.
thinning	Character; specification of the thinning operator. Currently, binomial ("binomial"), negative binomial ("nbinomial"), and Poisson ("poisson") thinning operators are available.
innovation	Character specification of the innovation process, see details.
n.start	The length of 'burn-in' period. If NA, the default, a reasonable value (500) is computed.

Details

There are some discrete distributions available for the innovation process specification. The following table display their names and their abbreviations to be passed to innovation().

Distribution	Abbreviation		Parameters
Bernoulli	"BE"		0 < theta < 1
BerPoi	"BP"		theta > 0; 0 < phi < 1
BerG	"BG"		theta, phi > 0
Geometric	"GE"		theta > 0
Mean-Parameterized COM-Poisson	"CP"		theta, phi > 0
Negative Binomial	"NB"		theta, phi > 0
Poisson	"PO"		theta > 0

Value

A time-series object of class "ts".

References

Du, J.G. and Li,Y. (1991). The integer-valued autorregressive (INAR(p)) model. Journal of time series analysis. 12, 129–142.

Examples

# A Poisson INAR(3) simulation
y <- inar.sim(n = 1000, alpha = c(0.2, 0.3, 0.3), par = 5)
layout(matrix(c(1,2,1,2,1,3,1,3), ncol = 4))
plot(y)
acf(y, main = "")
pacf(y, main = "")
layout(1)

rdmatheus/tsinteger documentation built on March 24, 2021, 12:16 a.m.