Libraries

Load libraries to complete this short tutorial.

# Time Series ML
library(tidymodels)
library(modeltime)
library(modeltime.ensemble)

# Core
library(tidyverse)
library(timetk)

interactive <- FALSE

Collect the Data

We’ll use the m750 dataset that comes with modeltime.ensemble. We can visualize the dataset.

m750 %>%
    plot_time_series(date, value, .color_var = id, .interactive = interactive)

getting-started-with-modeltime-ensemble.R

Recipe

We’ll create a Feature Engineering Recipe that can be applied to the data to create features that machine learning models can key in on. This will be most useful for the Elastic Net (Model 3).

recipe_spec <- recipe(value ~ date, training(splits)) %>%
    step_timeseries_signature(date) %>%
    step_rm(matches("(.iso$)|(.xts$)")) %>%
    step_normalize(matches("(index.num$)|(_year$)")) %>%
    step_dummy(all_nominal()) %>%
    step_fourier(date, K = 1, period = 12)

recipe_spec %>% prep() %>% juice()

#> # A tibble: 282 × 42
#>    date       value date_index.num date_year date_half date_quarter date_month
#>    <date>     <dbl>          <dbl>     <dbl>     <int>        <int>      <int>
#>  1 1990-01-01  6370          -1.72     -1.66         1            1          1
#>  2 1990-02-01  6430          -1.71     -1.66         1            1          2
#>  3 1990-03-01  6520          -1.70     -1.66         1            1          3
#>  4 1990-04-01  6580          -1.69     -1.66         1            2          4
#>  5 1990-05-01  6620          -1.67     -1.66         1            2          5
#>  6 1990-06-01  6690          -1.66     -1.66         1            2          6
#>  7 1990-07-01  6000          -1.65     -1.66         2            3          7
#>  8 1990-08-01  5450          -1.64     -1.66         2            3          8
#>  9 1990-09-01  6480          -1.62     -1.66         2            3          9
#> 10 1990-10-01  6820          -1.61     -1.66         2            4         10
#> # ℹ 272 more rows
#> # ℹ 35 more variables: date_day <int>, date_hour <int>, date_minute <int>,
#> #   date_second <int>, date_hour12 <int>, date_am.pm <int>, date_wday <int>,
#> #   date_mday <int>, date_qday <int>, date_yday <int>, date_mweek <int>,
#> #   date_week <int>, date_week2 <int>, date_week3 <int>, date_week4 <int>,
#> #   date_mday7 <int>, date_month.lbl_01 <dbl>, date_month.lbl_02 <dbl>,
#> #   date_month.lbl_03 <dbl>, date_month.lbl_04 <dbl>, …

getting-started-with-modeltime-ensemble.R

Model 1 - Auto ARIMA

First, we’ll make an ARIMA model using Auto ARIMA.

model_spec_arima <- arima_reg() %>%
    set_engine("auto_arima")

wflw_fit_arima <- workflow() %>%
    add_model(model_spec_arima) %>%
    add_recipe(recipe_spec %>% step_rm(all_predictors(), -date)) %>%
    fit(training(splits))

getting-started-with-modeltime-ensemble.R

Model 2 - Prophet

Next, we’ll make a Prophet Model.

model_spec_prophet <- prophet_reg() %>%
    set_engine("prophet")

wflw_fit_prophet <- workflow() %>%
    add_model(model_spec_prophet) %>%
    add_recipe(recipe_spec %>% step_rm(all_predictors(), -date)) %>%
    fit(training(splits))

getting-started-with-modeltime-ensemble.R

Model 3 - Elastic Net

Third, we’ll make an Elastic Net Model using glmnet.

model_spec_glmnet <- linear_reg(
    mixture = 0.9,
    penalty = 4.36e-6
) %>%
    set_engine("glmnet")

wflw_fit_glmnet <- workflow() %>%
    add_model(model_spec_glmnet) %>%
    add_recipe(recipe_spec %>% step_rm(date)) %>%
    fit(training(splits))

getting-started-with-modeltime-ensemble.R

Step 1 - Create a Modeltime Table

Create a Modeltime Table using the modeltime package.

m750_models <- modeltime_table(
    wflw_fit_arima,
    wflw_fit_prophet,
    wflw_fit_glmnet
)

m750_models

#> # Modeltime Table
#> # A tibble: 3 × 3
#>   .model_id .model     .model_desc            
#>       <int> <list>     <chr>                  
#> 1         1 <workflow> ARIMA(0,1,1)(0,1,1)[12]
#> 2         2 <workflow> PROPHET                
#> 3         3 <workflow> GLMNET

getting-started-with-modeltime-ensemble.R

Step 2 - Make an Ensemble

Then use ensemble_average() to turn that Modeltime Table into a Modeltime Ensemble. This is a fitted ensemble specification containing the ingredients to forecast future data and be refitted on data sets using the 3 submodels.

ensemble_fit <- m750_models %>%
    ensemble_average(type = "mean")

ensemble_fit

#> ── Modeltime Ensemble ───────────────────────────────────────────
#> Ensemble of 3 Models (MEAN) 
#> 
#> # Modeltime Table
#> # A tibble: 3 × 3
#>   .model_id .model     .model_desc            
#>       <int> <list>     <chr>                  
#> 1         1 <workflow> ARIMA(0,1,1)(0,1,1)[12]
#> 2         2 <workflow> PROPHET                
#> 3         3 <workflow> GLMNET

getting-started-with-modeltime-ensemble.R

Step 3 - Forecast! (the Test Data)

To forecast, just follow the Modeltime Workflow.

# Calibration
calibration_tbl <- modeltime_table(
    ensemble_fit
) %>%
    modeltime_calibrate(testing(m750_splits))

# Forecast vs Test Set
calibration_tbl %>%
    modeltime_forecast(
        new_data    = testing(m750_splits),
        actual_data = m750
    ) %>%
    plot_modeltime_forecast(.interactive = interactive)

getting-started-with-modeltime-ensemble.R

Step 4 - Refit on Full Data & Forecast Future

Once satisfied with our ensemble model, we can modeltime_refit() on the full data set and forecast forward gaining the confidence intervals in the process.

refit_tbl <- calibration_tbl %>%
    modeltime_refit(m750)

refit_tbl %>%
    modeltime_forecast(
        h = "2 years",
        actual_data = m750
    ) %>%
    plot_modeltime_forecast(.interactive = interactive)

getting-started-with-modeltime-ensemble.R

This was a very short tutorial on the simplest type of forecasting, but there’s a lot more to learn.

Take the High-Performance Forecasting Course

Become the forecasting expert for your organization

High-Performance Time Series Course