Error in ETS forecast

[robjhyndman]

As reported at https://github.com/orgs/OTexts/discussions/31

library(fpp3)
#> Registered S3 method overwritten by 'tsibble':
#>   method               from 
#>   as_tibble.grouped_df dplyr
#> ── Attaching packages ──────────────────────────────────────────── fpp3 1.0.1 ──
#> ✔ tibble      3.2.1     ✔ tsibble     1.1.6
#> ✔ dplyr       1.1.4     ✔ tsibbledata 0.4.1
#> ✔ tidyr       1.3.1     ✔ feasts      0.4.1
#> ✔ lubridate   1.9.4     ✔ fable       0.4.1
#> ✔ ggplot2     3.5.2
#> ── Conflicts ───────────────────────────────────────────────── fpp3_conflicts ──
#> ✖ lubridate::date()    masks base::date()
#> ✖ dplyr::filter()      masks stats::filter()
#> ✖ tsibble::intersect() masks base::intersect()
#> ✖ tsibble::interval()  masks lubridate::interval()
#> ✖ dplyr::lag()         masks stats::lag()
#> ✖ tsibble::setdiff()   masks base::setdiff()
#> ✖ tsibble::union()     masks base::union()

train <- data.frame(
    YM = seq(yearmonth(ymd("20220101")), yearmonth(ymd("20241201")), by = 1),
    value = c(rep(-78040, 11), -78061, rep(-78040, 11), -78061, rep(-78040, 11), -78061)
) %>% 
    as_tsibble(index = YM)

fit <- train %>%
    model(ETS(value))

fit %>%
    forecast(h = 12)  # Error
#> Error in `mutate()`:
#> ℹ In argument: `ETS(value) = (function (object, ...) ...`.
#> Caused by error in `[<-`:
#> ! only 0's may be mixed with negative subscripts

^{Created on 2025-05-20 with reprex v2.1.1}

[mitchelloharawild]

This is an obscure bug in the automatic selection process, where -Inf ic values caused the selection criteria for the 'best' model to be different from the model actually chosen.

Fixed.

library(fable)
#> Loading required package: fabletools
library(tsibble)
#> 
#> Attaching package: 'tsibble'
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, union
train <- tsibble(
  YM = yearmonth("2022 Jan") + 0:35,
  value = rep(c(rep(-78040, 11), -78061), 3),
  index = YM
)

fit <- train |> 
  model(ets=ETS(value))

fit %>%
  forecast(h = 12) 
#> # A fable: 12 x 4 [1M]
#> # Key:     .model [1]
#>    .model       YM
#>    <chr>     <mth>
#>  1 ets    2025 Jan
#>  2 ets    2025 Feb
#>  3 ets    2025 Mar
#>  4 ets    2025 Apr
#>  5 ets    2025 May
#>  6 ets    2025 Jun
#>  7 ets    2025 Jul
#>  8 ets    2025 Aug
#>  9 ets    2025 Sep
#> 10 ets    2025 Oct
#> 11 ets    2025 Nov
#> 12 ets    2025 Dec
#> # ℹ 2 more variables: value <dist>, .mean <dbl>

^{Created on 2025-05-20 with reprex v2.1.1}