This repo provides two R packages; see individual README.md files for more information. Briefly,
-
psSLP: This is a general use package that provides functions aiding in combining quantile functions or modeling/forecasts outputs. Includes:
- functions to pre-smoothing longitudinal sets of quantile estimations
- functions for generating both simulated linear pooling and vincentization ensemble estimates
- is intended for pre-smoothing and simulating tens of thousands of forecasts, and by default is parallelized
Installation:
devtools::install_github("lmullany/JHUAPL-SLPHospEns",subdir="psSLP") -
covid19hosp: This is a context/task-specific, non-general wrapper package for preparing data inputs to psSLP from data available at the Covid 19 Forecasting Hub (https://github.com/reichlab/covid19-forecast-hub).
- Requires local access to the hub data, preferably through a clone of the repo
- Provides functions for pulling modeling group specific quantile functions
- Includes checks to examine data for (longitudinal) consistencies
- Built-in functions for pulling empirical data from healthdata.gov
- Aids in estimating coverage of ensembles
- Includes plotting functions to aid in displaying results
Installation:
devtools::install_github("lmullany/JHUAPL-SLPHospEns",subdir="covid19hosp")
This work was supported by the US Department of Health and Human Services Office of the Assistant Secretary of Preparedness and Response