Raw output is saved to disk in the raw_output_dir specified in the eval config
as serialized .rds files. Processed output is saved in a structured directory
format in the output_dir specified in the eval config. See the save_table(),
to_rds_with_suffix(), and get_raw_output_suffix() functions for more details.
Usage
eval_postprocess(
forecast_date,
eval_date,
location,
model,
scenario,
hosp_data_dir,
ww_data_dir,
ww_data_mapping,
scenario_dir,
output_dir,
raw_output_dir,
max_eval_data_days = 365
)Arguments
- forecast_date
As-of date for the forecast.
- eval_date
As-of date for evaluation data to use.
- location
Location to forecast.
- model
Model to postprocess. One of
"ww"and"hosp"- scenario
Wastewater data availability scenario to analyze.
- hosp_data_dir
Path to a directory containing vintaged hospital admissions data in date-stamped .csv files.
- ww_data_dir
Path to a directory containing vintaged wastewater data in date-stamped .csv files.
- ww_data_mapping
String associating forecast dates to wastewater vintage dates. Passed to
date_of_ww_data().- scenario_dir
Path to a directory containing .csv files that define wastewater data availability scenarios.
- output_dir
Path to a direcotry in which to save general postprocess output.
- raw_output_dir
Path to a directory in which to save raw output as serialized
.rdsfiles.- max_eval_data_days
Maximum number of days of data to pull when creating evaluation dataset. Default 365.
Details
The bulk of the postprocessing for successful model fits is handled
by the postprocess_successful_fit() function, which is called within this
function provided the target model fit was indeed successful, but there are
some postprocessing tasks we wish to perform regardless of
whether the model fit was successful.