Currently, we are using the post-processed outputs (ww quantiles joined to input data) to identify quality issues (high latency, extremely high noise, basically no sites, etc) where we would want to exclude that location-forecast date from the head-to-head comparison. This is not great, bc theres tons of wastewater data and its unnecessarily slow, and these flags are only a function of the input data
Requirements
[ ] modify command_line_post_process() and associated functions to compute an individual state-forecast-date's ww flags to determine if the wastewater data is sufficient
[ ] modify command_line_post_process() and associated functions to save the input data for reproducibilty (both ww and hosp)
[ ] modify _targets_eval_postprocessing.R() to pull in all of the ww_data_flags from across the nested file structure and get the table of sufficient ww data. Use this instead of the current workflow, which relies on the full set of quantiles and is therefore very slow
[ ] only pull in the ww quantiles needed for plotting Fig 2, we don't need to hold all of them in memory
Goal
Currently, we are using the post-processed outputs (ww quantiles joined to input data) to identify quality issues (high latency, extremely high noise, basically no sites, etc) where we would want to exclude that location-forecast date from the head-to-head comparison. This is not great, bc theres tons of wastewater data and its unnecessarily slow, and these flags are only a function of the input data
Requirements
command_line_post_process()and associated functions to compute an individual state-forecast-date's ww flags to determine if the wastewater data is sufficientcommand_line_post_process()and associated functions to save the input data for reproducibilty (both ww and hosp)_targets_eval_postprocessing.R()to pull in all of theww_data_flagsfrom across the nested file structure and get the table of sufficient ww data. Use this instead of the current workflow, which relies on the full set of quantiles and is therefore very slow