Skip to content

Latest commit

 

History

History
45 lines (32 loc) · 2.64 KB

README.md

File metadata and controls

45 lines (32 loc) · 2.64 KB
Raw

Evaluating an epidemiologically motivated surrogate model of a multi-model ensemble

This repository contains the documentation, results, and code of a project evaluating a simplified forecasting model in comparison the European forecasting hub ensemble. See the documentation for further details.

Citation

Please cite this using the following:

Abbott, Sherratt, Bosse, Gruson, Bracher, and Funk. 2022 “Evaluating an Epidemiologically Motivated Surrogate Model of a Multi-Model Ensemble.” medRxiv. https://doi.org/10.1101/2022.10.12.22280917.

@UNPUBLISHED{Abbott_undated-tu,
  title   = "Evaluating an epidemiologically motivated surrogate model of a
             multi-model ensemble",
  author  = "{Abbott} and {Sherratt} and {Bosse} and {Gruson} and {Bracher} and
             {Funk}",
  journal = "medRxiv",
  year    = 2022,
  doi     = "10.1101/2022.10.12.22280917"
}

Project structure

Folder Purpose
ecdc-weekly-growth-forecasts The code for the simplified forecasting model evaluated in this work.
data-raw Raw input data and scripts required to download and process it.
data Processed data from data-raw ready to be used in the paper analysis.
R R functions used in the analysis and for evaluation.
paper Summary paper and additional supplementary information as Rmarkdown documents.
.github GitHub actions used to build the docker image and render and publish the analysis paper.
.devcontainer Resources for reproducibility using vscode and docker.

Dependencies

Dependencies are managed using renv.

Alternatively a docker container and image is provided. An easy way to make use of this is using the Remote development extension of vscode.

Reproducibility

Once all dependencies are installed (see above) the paper analysis can be rerun using paper/paper.Rmd either interactively or rerendered as a document using Rmarkdown. To make this step easier we also provide a GitHub action to publish an updated version of the analysis to the gh-pages branch.

See data-raw for the code to re-extract forecasts and truth data, create metadata, normalise by population, and score forecasts against truth data. All steps of this process can be done automatically using data-raw/update.sh. Results from these steps will be stored in data as .csv files.