All models failed Modeltime Calibration:

[nomanres]

I have a set of equal univariate timeseries with id and prepared recipe like this:

rec_obj <- recipe(value ~ ., training(splits)) %>%
step_mutate_at(id, fn = droplevels) %>%
step_date(date, features = c("month", "quarter", "year")) %>%
step_mutate(dt = as.numeric(date)) %>%
step_rm(date) %>%
step_dummy(all_nominal_predictors(), one_hot = TRUE)

And I use code from Forecasting with Global Models (modeltime
1.2.8.9000) with xgboost model.
I cannot calibrate my model
calib_tbl <- model_tbl %>%
modeltime_calibrate(
new_data = testing(splits),
id = "id"
)

calib_tbl

[nomanres]

using step_timeseries_signature() gives the same error

[mdancho84]

Without a full reproducible example, I'm not able to help.

Ideas: Try updating your R packages.