Measuring Contributions of Component Models to Ensemble Forecast Accuracy

Provides metrics for quantifying the contribution of individual component models to the predictive accuracy of ensemble forecasts. The package implements the Leave-One-Model-Out (LOMO) and Leave-All-Subset-of-One-Model-Out (LASOMO) model importance metrics, enabling users to assess the relative importance of component models and better understand the performance of ensemble forecasting systems. Methods are described in Kim et al. (2026) .

modelimportance

Overview

The goal of modelimportance is to provide tools to quantify each individual model’s contribution to an ensemble model’s predictive performance. Importance scores for each ensemble member are computed based on their impact on the ensemble’s accuracy, helping users understand which models are most influential in improving the ensemble’s predictions. The package is designed to work with the standard S3 class model output format defined by the hubverse convention.

Installation

You can install the development version of modelimportance from GitHub with:

remotes::install_github("mkim425/modelimportance")

Getting started

library(modelimportance)

The main function in the package is model_importance(), which calculates the importance score for each model in an ensemble for individual prediction tasks. The output of model_importance() is an S3 object of class model_imp_tbl. This object can be further analyzed using various methods such as print(), summary(), and aggregate(), which offer different ways to interpret the importance scores.

Learn more

Learn more about this package and how to use it in the vignette. The vignette provides detailed examples and theoretical background on the algorithms implemented in the package.

Code of Conduct

Please note that this project is released with a Code of Conduct. By participating in this project, you agree to abide by its terms.