Get plot of overall hub performance, grouped by period

Usage

make_fig5_hub_performance(
  all_scores,
  cfa_real_time_scores,
  figure_file_path,
  all_time_period,
  real_time_period,
  models_to_show,
  summarize_across_horizon = FALSE,
  baseline_model = "COVIDhub-4_week_ensemble"
)

Arguments

all_scores: df with granular (daily) scores from every model, forecast_date, and location for the entire time period. Includes the two retrospective models
cfa_real_time_scores: df with granular (daily) scores from the submitted cfa ww model for the time period when it was submitted.
figure_file_path: path to directory to save figures
all_time_period: string indicating the longer time frame we are comparing, e.g. "Oct 2023-Mar 2024"
real_time_period: string indicating the shorter time frame that we submitted our model to the hub e.g. "Feb 2024-Mar 2024"
models_to_show: A vector of charcter strings indicating which models from the COVID-19 forecast hub to include in the plot.
summarize_across_horizon: Boolean indicating whether or not to average the scores across the horizon, default is FALSE meaning each day-forecast-date-location score is in the distribution
baseline_model: which model to compute relative WIS compared to, default is COVIDhub-4_week_ensemble

Value

a ggplot object containing distributions of WIS scores grouped by model and the comaprison time period, with the mean plotted alongside the full distribution