TSA (version 1.3)

plot.Arima: Compute and Plot the Forecasts Based on a Fitted Time Series Model

Description

Plots the time series data and its predictions with 95% prediction bounds.

Usage

# S3 method for Arima
plot(x, n.ahead = 12, col = "black", ylab = object$series, 
lty = 2, n1, newxreg, transform, Plot=TRUE, ...)

Arguments

x

a fitted arima model

n.ahead

number of prediction steps ahead (default=12)

col

color of the prediction bounds

ylab

label of the y-axis

lty

line type of the point predictor; default=dashed lines

n1

starting time point of the plot (default=earliest time point)

newxreg

a matrix of covariate(s) over the period of prediction

transform

function used to transform the forecasts and their prediction bounds; if missing, no transformation will be carried out. This option is useful if the model was fitted to the transformed data and it is desirable to obtain the forecasts on the original scale. For example, if the model was fitted with the logarithm of the data, then transform = exp will plot the forecasts and their prediction bounds on the original scale.

Plot

Plotting will be suppressed if Plot is set to be FALSE; default is TRUE

…

additional parameters passed to the plot function

Value

Side effects of the function: plot the forecasts and their 95% prediction bounds, unless Plot is set to be FALSE. The part of the observed series is plotted with all data plotted as open circles and linked by a smooth line. By default the predicted values are plotted as open circles joined up by a dashed line. The plotting style of the predicted values can be altered by supplying relevant plotting options, e.g specifying the options type='o', pch=19 and lty=1 will plot the predicted values as solid circles that are overlaid on the connecting smooth solid line. The prediction limits are plotted as dotted lines, with default color being black. However, the prediction limits can be drawn in other colors. For example, setting col='red' paints the prediction limits in red. An interesting use of the col argument is setting col=NULL which has the effect of not drawing the prediction limits.

The function returns an invisible list containing the following components.

pred

the time series of predicted values

lpi

the corresponding lower 95% prediction limits

upi

the corresponding upper 95% prediction limits

Examples

# NOT RUN {
data(oil.price)
oil.IMA11alt=arima(log(oil.price),order=c(0,1,1),
# create the design matrix of the covariate for prediction
xreg=data.frame (constant=seq(oil.price)))
n=length(oil.price)
n.ahead=24
newxreg=data.frame(constant=(n+1):(n+n.ahead))
# do the prediction and plot the results
plot(oil.IMA11alt,n.ahead=n.ahead,newxreg=newxreg,
ylab='Log(Oil Price)',xlab='Year',n1=c(2000,1))
# do the same thing but on the orginal scale
plot(oil.IMA11alt,n.ahead=n.ahead,newxreg=newxreg,
ylab='Oil Price',xlab='Year',n1=c(2000,1),transform=exp,pch=19, lty=1,type='o')
# Setting pch=19 plots the predicted values as solid circles.
res=plot(oil.IMA11alt,n.ahead=n.ahead,newxreg=newxreg,
ylab='Oil Price',xlab='Year',n1=c(2000,1),transform=exp,pch=19,col=NULL)
# Setting col=NULL will make the prediction bands invisible. Try col='red'. 
res
# prints the predicted values and their 95% prediction limits.

# }