FitAR (version 1.94)

VarianceRacfARp: Covariance Matrix Residual Autocorrelations for ARp

Description

The ARp subset model is defined by taking a subset of the parameters in the regular AR(p) model. With this function one can obtain the standard deviations of the residual autocorrelations which can be used for diagnostic checking with RacfPlot.

Usage

VarianceRacfARp(phi, lags, MaxLag, n)

Arguments

phi

vector of AR coefficients

lags

lags in subset AR

MaxLag

covariance matrix for residual autocorrelations at lags 1,...,m, where m=MaxLag is computes

n

length of time series

Value

The m-by-m covariance matrix of residual autocorrelations at lags 1,...,m, where m = MaxLag.

Details

The covariance matrix for the residual autocorrelations is derived in McLeod (1978, eqn. 15) for the general ARMA case. McLeod (1978, eqn. 35) specializes this result to the subset case.

References

McLeod, A.I. (1978), On the distribution and applications of residual autocorrelations in Box-Jenkins modelling, Journal of the Royal Statistical Society B, 40, 296-302.

See Also

VarianceRacfAR, VarianceRacfARz, RacfPlot

Examples

#the standard deviations of the first 5 residual autocorrelations
#to a subset AR(1,2,6) model fitted to Series A is
v<-VarianceRacfARp(c(0.36,0.23,0.23),c(1,2,6), 5, 197)
sqrt(diag(v))