CODEWITHSUNDEEP
rtime-serieshierarchicalforecasting
Selcuk Akbas
Selcuk Akbas

Reputation: 711

what is the arima parameters of a hts hierarchical or grouped time series forecast?

Is there any way to find the arima parameters for a hts forecast ?

My forecast is something like this:

myts_f <- forecast(myts, h=78, fmethod = "arima", method = "tdfp")

hts is: http://cran.r-project.org/web/packages/hts

Thanks, Regards

Upvotes: 2

Views: 911

Answers (1)

Rob Hyndman
Rob Hyndman

Reputation: 31800

A separate ARIMA model is estimated for every series in the hierarchy. For large hierarchies, that can involve thousands or even millions of models. There is no point in storing all the resulting information when you only want forecasts.

If you really care about the individual models, then fit them explicitly to all series. You can get the matrix of every series (included aggregated series) using aggts(myts). For example:

y <- aggts(myts)
models <- list()
for(i in 1:ncol(y))
  models[[i]] <- auto.arima(y[,i])

Upvotes: 2

Related Questions