-
Follow up to my comments on #316.
When using `h>1` to do h-step forecasts, results are all `NA`. For example:
```
fit1
-
was there a reason we didn't include `thetaf` in the ensemble, even as an option? If we just didn't get around to it, I'm happy to put some belated work in to add it.
-
This should be pretty easy to implement like these sources:
http://robjhyndman.com/hyndsight/tscvexample/
http://www.r-bloggers.com/functional-and-parallel-time-series-cross-validation/
The biggest t…
-
The Arima command allows for a constant, while the auto.arima does not seem so.
In the example below the comovement between y and x is very weak but auto.arima forces the regression through the origi…
-
I think the auto.arima function is bugged, the D parameter doesn't change the result:
```
library(forecast)
auto.arima(WWWusage, D=0)
auto.arima(WWWusage, D=1)
auto.arima(WWWusage, D=2)
```
All giv…
-
`forecast.xgbts()` throws a warning if no `h` is provided and defaults to 24. You might want to save the frequency of the input time series in the `xgbts` object and default to `2 * frequency(inputSer…
-
I had the install multiple dependencies manually:
- fracdiff
- tseries
- timeDate
You should add these to the dependencies if they are necessary. Otherwise, users trying to install get errors like:
…
-
1. For es() - when number of observations is not enough to fit ZZZ, make a pool of models with ZNN, ZZN, ZNZ - depending on the data frequency and sample size.
2. For auto.arima() - restrict orders de…
-
See below
```
hm2
-
Dear Rob,
I am using Package Forecast, function auto.arima with the following command:
ARIMA2_dummy