Hello,
I have a general question, I am fitting and validating a seasonal arima
model on daily time series data, the data have an obvious (7 day weekly)
seasonal trend using an arima (1,0,1)x(0,1,1). I have a validation set of
thirteen weeks and I have computed one day ahead forecasts, i.e. I
incrementally add one day from the validation set to the test set
reestimate the model parameters and then make a forecast for the next day in
the validation set, I do the same thing for 7 days ahead, i.e. adding in
increments of 7 reestimating and then forecasting for the following 7 days,
and then for 30 days ahead and for the full 91 days ahead. When I computed
root mean squared error (rmse) for the forecasts at different horizons I was
surprised that the rmse's were lower for the 30 day horizons and 91 day
horizon than they were for both the 7-day horizon and the 1-day ahead
forecasts. I have tried to understand why this might be happening,but I
would appreciate any feed back if anyone has the time. Sorry for the general
question
best,
Spencer Jones
[[alternative HTML version deleted]]
______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
