2

I'm working with multiple time series models, and hoping to manage them with data.table:

pkg <- c("data.table", "magrittr", "forecast")
sapply(pkg, library, character.only = TRUE)

df <- data.frame(
  group = rep(c("a", "b", "c"), each = 10),
  val = sample(1:10, 30, replace = TRUE)
) %>% as.data.table

I can successfully generate the model (and a time series ready for the next step):

t1 <- df[, list(
  tsAll = list(val %>% as.ts),
  mod1 = list(val %>% as.ts %>% window(1, 7) %>% ets)
  ), by = group]

I'm now trying to update the ets model, which involves passing 'tsAll' into the ets function and providing the model to use (mod1).

This does not work:

t1[, lapply(tsAll, ets, model = mod1)]

I've also tried:

 t1[, lapply(tsAll, ets, model = mod1[[1]])]

This runs, but it looks like the same model is returned to every row.

Possibly I have solved the question but run into this issue:

Why is using update on a lm inside a grouped data.table losing its model data?

Can anyone help with the next step?

Community
  • 1
  • 1
Ed G
  • 802
  • 1
  • 6
  • 19

1 Answers1

1

Is this what you want:

t1[, list(lapply(tsAll, ets, model = mod1[[1]])), by = group]$V1

I put the result in a list, so that the data type is preserved, as opposed to being converted into a vector and did the operation by group (since each group has its own model).

eddi
  • 49,088
  • 6
  • 104
  • 155
  • That's it - so simple when you see it - thanks. I've been getting an intermittent error where the first model is returned to all rows. If I can pin down exactly what caused it I'll update the question. – Ed G Feb 12 '15 at 12:56