R Graphical Manual

Browse All

Last data update: 2014.03.03

R: Model Diagnostics for a Fitted ARIMA Model

tsdiag.Arima

R Documentation

Model Diagnostics for a Fitted ARIMA Model

Description

This function is modified from the tsdiag function of the stats package.

Usage

## S3 method for class 'Arima'
tsdiag(object, gof.lag, tol = 0.1, col = "red", omit.initial = TRUE,...)

Arguments

`object`	a fitted ARIMA model
`gof.lag`	maximum lag used in ACF and Ljung-Box tests for the residuals
`tol`	tolerance (default=0.1); see below
`col`	color of some warning lines in the figures (default=red)
`omit.initial`	suppress the initial (d+Ds) residuals if true
`...`	other arguments to be passed to the acf function

Value

Side effects: Plot the time plot of the standardized residuals. Red dashed line at +/-qnorm(0.025/no of data) are added to the plot. Data beyond these lines are deemed outliers, based on the Bonferronni rule. The ACF of the standardized residuals is plotted. The p-values of the Ljung-Box test are plotted using a variety of the first K residuals. K starts at the lag on and beyond which the moving-average weights (in the MA(infinity) representation) are less than tol.

Author(s)

Kung-Sik Chan, based on the tsdiag function of the stats pacakage

Examples

data(color)
m1.color=arima(color,order=c(1,0,0))
tsdiag(m1.color,gof=15,omit.initial=FALSE)