predict.artfima {artfima}

Predict method for artfima

Description

The optimal minimum mean square error forecast and its standard deviation for lags 1, 2, ..., n.ahead is computed at forecast origin starting at the end of the observed series used in fitting. The exact algorithm discussed in McLeod, Yu and Krougly is used.

Usage

## S3 method for class 'artfima'
predict(object, n.ahead=10, ...)

Arguments

object

object of class "artfima"

n.ahead

number of steps ahead to forecast

...

optional arguments

Value

a list with two components

Forecasts

Description of 'comp1'

SDForecasts

Description of 'comp2'

Author(s)

A. I. McLeod, [e-mail]

References

McLeod, A.I., Yu, Hao and Krougly, Z. (2007). Algorithms for Linear Time Series Analysis: With R Package. Journal of Statistical Software 23/5 1-26.

See Also

predict.Arima

Examples

ans <- artfima(seriesa, likAlg="Whittle")
predict(ans)
#compare forecasts from ARTFIMA etc.
  ## Not run: 
ML <- 10
ans <- artfima(seriesa)
Ftfd <- predict(ans, n.ahead=10)$Forecasts 
ans <- artfima(seriesa, glp="ARIMA", arimaOrder=c(1,0,1))
Farma11 <- predict(ans, n.ahead=10)$Forecasts 
ans <- artfima(seriesa, glp="ARFIMA")
Ffd <- predict(ans, n.ahead=10)$Forecasts
#arima(0,1,1)
ans <- arima(seriesa, order=c(0,1,1))
fEWMA <- predict(ans, n.ahead=10)$pred
yobs<-seriesa[188:197]
xobs<-188:197
y <- matrix(c(yobs,Ffd,Ftfd,Farma11,fEWMA), ncol=5)
colnames(y)<-c("obs", "FD", "TFD", "ARMA11","FEWMA")
x <- 197+1:ML
x <- matrix(c(xobs, rep(x, 4)), ncol=5)
plot(x, y, type="n", col=c("black", "red", "blue", "magenta"),
     xlab="t", ylab=expression(z[t]))
x <- 197+1:ML
points(xobs, yobs, type="o", col="black")
points(x, Ffd, type="o", col="red")
points(x, Ftfd, type="o", col="blue")
points(x, Farma11, type="o", col="brown")
points(x, fEWMA, type="o", col="magenta")
legend(200, 18.1, legend=c("observed", "EWMA", "FD", "TFD", "ARMA"),
       col=c("black", "magenta", "red", "blue", "brown"),
       lty=c(rep(1,5)))
  
## End(Not run)