How to obtain the Confidence Intervals

[mbanco]

Hello Matt,

I don't know how to get the confidence intervals in the projection:

library(modeltime);
library(tidymodels);
library(workflowsets);
library(tidyverse);
library(timetk);

nivel_confianza <- 0.75;
cant_proyec <- 60;

dataset_tbl <- walmart_sales_weekly %>% select(id, Date, Weekly_Sales) %>% set_names(c('id', 'fecha', 'datos'));

cant <- dataset_tbl %>% group_by(id) %>% count(id) ;

splits <- time_series_split(
dataset_tbl,
date_var = fecha,
assess = ceiling(min(cant$n) * 0.3),
cumulative = TRUE
);

splits %>%

tk_time_series_cv_plan() %>%

plot_time_series_cv_plan(fecha, datos, .interactive = F);

recipe_spec <- recipe(datos ~ ., data = training(splits)) %>%
step_timeseries_signature(fecha) %>%
step_rm(fecha) %>%
step_normalize(fecha_index.num) %>%
step_zv(all_predictors()) %>%
step_dummy(all_nominal_predictors(), one_hot = TRUE);

model_tbl <- tibble(
learn_rate = c(0.001, 0.010, 0.100, 0.350, 0.500, 0.650)) %>%
create_model_grid(f_model_spec=boost_tree, engine_name="xgboost", mode="regression");

model_list <- model_tbl$.models;

model_wfset <- workflow_set(
preproc = list(
recipe_spec
),
models = model_list,
cross = TRUE
);

model_parallel_tbl <- model_wfset %>%
modeltime_fit_workflowset(
data = training(splits),
control = control_fit_workflowset(
verbose = TRUE,
allow_par = TRUE
)
);

test <- model_parallel_tbl %>%
modeltime_calibrate(testing(splits), id="id", conf_interval=nivel_confianza, conf_by_id=TRUE) %>%
modeltime_accuracy(acc_by_id = FALSE);

refit_tbl <- model_parallel_tbl %>%
modeltime_refit(data = dataset_tbl, control_refit(allow_par=TRUE));

future_tbl <- dataset_tbl %>%
group_by(id) %>%
future_frame(.length_out=cant_proyec, .bind_data=FALSE, .date_var=fecha);

projection <- refit_tbl[which.min(test$rmse),] %>%
modeltime_forecast(
new_data = future_tbl,
actual_data = dataset_tbl,
keep_data = FALSE,
conf_interval = nivel_confianza,
conf_by_id = TRUE
) %>%
group_by(id);

parallel_stop();

Thanks.

Best Regards

Mauricio