Simple models for COVID-19 hospitalizations

The models built here use the HHS Protect daily COVID-19 hospital admissions data.

We create a set of simple baseline models that project existing trends into the future. Each baseline has a different set of options specified. We then combine the baseline models into a single ensemble forecast.

setup

1. Install the simplets R package ("simple time-series" models).
2. Install the epitools R package (as of Jan 6, 2022, you need to install a specific branch on a specific fork).
3. Install the hubEnsembles R package.
4. Install the covidHubUtils R package.
5. Clone this repo and the flu-hosp-models-2021-2022 repo so both clones live in the same home directory.

workflow

1. run make all in root directory
2. submit forecast submission file from weekly-submission/forecasts/UMass-trends_ensemble/ to the COVID-19 Forecast Hub as a PR.
3. commit and push generated CSV files from all models and the PDF of plots from JUST the trends_ensemble model.