Update: the two problems described below each trace to a misspecified coercion in as_tibble
I had
obj %>% as_tibble(key = c(X,Y))
Changing it to
obj %>% as_tibble(index = Month)
eliminated the error messages.
I've come to Chapter 7 in the fpp3 book on forecasting and become stuck at series %>% model(TSLM(Y ~ X))
, where series is a tsibble
with Month an mth
, X, Y, X & Y both doubles
(in different units) and Y >> X.
Possibly related issue
Error in FUN(X[[i]], ...) : object 'kwh' not found
when using model
I can't see any difference between this and the examples, which do work.
olympic_running %>%
model(TSLM(Time ~ trend())) %>%
interpolate(olympic_running)
Everything is CRAN versioned, running under R 4.0.2 on Ubuntu 20.04 with
- fable_0.2.1
- feasts_0.1.5
- fabletools_0.2.1
- tsibbledata_0.2.0
- tsibble_0.9.3
BTW: the new package is a feast
of improved and more sophisticated tools over forecast