Timeline for auto.arima from Forecast package
Current License: CC BY-SA 3.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 14, 2013 at 13:09 | vote | accept | Ludo | ||
| Jan 14, 2013 at 13:09 | comment | added | Ludo | I was clearly comparing the AIC in the final output with the one in the trace. Thank you for the explanation. | |
| Jan 14, 2013 at 12:58 | comment | added | Rob Hyndman | The first model returns an ARIMA(2,0,2)(0,1,2) model with an AIC of 8183.52. The AIC returned by trace (using an approximation) is 7996.88. The second model gives the same AIC in the trace using the same approximation. Perhaps you are comparing the AIC returned with that printed in the trace. The latter is an approximation. If you want the exact AIC, use the argument approximation=FALSE. | |
| Jan 14, 2013 at 12:33 | comment | added | Ludo | Thank you for your reply. Unfortunately, I am not authorized to publish this data on the internet. However, if I make no mistake, I think I am able to replicate the problem using the gas time series provided in the Forecast package. Please consider the code : data <- gas xreg_past <- 1:length(data) auto.arima(data,d=0,D=1,max.p=2,max.q=2,max.P=2,max.Q=2,max.order=8, xreg=xreg_past,trace=TRUE,ic="aic") auto.arima(data,d=0,D=1,max.p=2,max.q=2,max.P=2,max.Q=2,max.order=8, xreg=xreg_past,stepwise=FALSE,trace=TRUE,ic="aic") | |
| Jan 12, 2013 at 7:08 | history | answered | Rob Hyndman | CC BY-SA 3.0 |