ensemble_weighted: Creates a Weighted Ensemble Model

In modeltime.ensemble: Ensemble Algorithms for Time Series Forecasting with Modeltime

Creates a Weighted Ensemble Model

Description

Makes an ensemble by applying loadings to weight sub-model predictions

Usage

ensemble_weighted(object, loadings, scale_loadings = TRUE)

Arguments

object	A Modeltime Table
loadings	A vector of weights corresponding to the loadings
scale_loadings	If TRUE, divides by the sum of the loadings to proportionally weight the submodels.

Details

The input to an ensemble_weighted() model is always a Modeltime Table, which contains the models that you will ensemble.

Weighting Method

The weighted method uses uses loadings by applying a loading x model prediction for each submodel.

Value

A mdl_time_ensemble object.

Examples

library(tidymodels)
library(modeltime)
library(modeltime.ensemble)
library(dplyr)
library(timetk)

# Make an ensemble from a Modeltime Table
ensemble_fit <- m750_models %>%
    ensemble_weighted(
        loadings = c(3, 3, 1),
        scale_loadings = TRUE
    )

ensemble_fit

# Forecast with the Ensemble
modeltime_table(
    ensemble_fit
) %>%
    modeltime_forecast(
        new_data    = testing(m750_splits),
        actual_data = m750
    ) %>%
    plot_modeltime_forecast(
        .interactive = FALSE,
        .conf_interval_show = FALSE
    )

modeltime.ensemble documentation built on Sept. 11, 2024, 6:42 p.m.