The output of clean_anomalies() is a time tibble, which doesn't play nice with timetk::future_frame()
mabuimo opened this issue · comments
Manuel Buitrago commented
Hi,
Let's clean the anomalies of a given dataset:
data(tidyverse_cran_downloads)
cleaned <- tidyverse_cran_downloads |>
anomalize::time_decompose(count, method = "stl") |>
anomalize::anomalize(remainder, method = "iqr") |>
anomalize::clean_anomalies()
The output is a time tibble. Now if you extend these time series by 12 points:
cleaned |>
dplyr::group_by(package) |>
timetk::future_frame(
.date_var = date ,
.length_out = 12,
.bind_data = FALSE
) |>
dplyr::ungroup()
You get:
Error in `tbl_at_vars()`:
! Can't subset columns that don't exist.
✖ Column `package` doesn't exist.
This error can be solved by ensuring that cleaned
is a tibble before extending the dataframe:
cleaned |>
as_tibble() |>
dplyr::group_by(package) |>
timetk::future_frame(
.date_var = date ,
.length_out = 12,
.bind_data = FALSE
) |>
dplyr::ungroup()
Thank you Matt!