Renders prediction interval ribbons plus the median as a ggplot.
Works on any hubverse quantile forecast, whether the values are
admissions or census. Any column not in the hubverse value triplet
(target_end_date, horizon, output_type*, value) is treated
as a trajectory grouper, so user added columns like los need no
configuration.
Usage
plot_fan(fcast, location = NULL, truth = NULL, intervals = c(0.5, 0.95))Arguments
- fcast
Hubverse format quantile tibble. Typically the output of
fcast_census(), but any admission or census quantile forecast in hubverse shape works.- location
Optional character. If supplied, both
fcastandtruthare filtered to this location before plotting.- truth
Optional hubverse target data tibble (columns
target_end_date,target,location,observation). Overlaid as a black line.- intervals
Numeric vector of prediction interval widths in
(0, 1). Defaultc(0.5, 0.95). One ribbon per interval.