The function estimate_input_cost_of_scenarios expects number_of_hcw_staff to logged so that column names look like this: FacilityID_0_Officer_Clinical etc.
Within get_staff_count_by_facid_and_officer_type FacilityID_0_Officer_Clinical is changed to FacilityID=0|Officer=Clinical, which is them passed to unflatten_flattened_multi_index_in_logging.
With the clinics implementation, number_of_hcw_staff is now logged so that each clinic constitutes a column, and the staff count is a dictionary. This change has broken the estimate_input_cost_of_scenarios.
I scanned the scripts directory, and it seems the key is being used only by estimate_input_cost_of_scenarios, but it might affect other analysis files that are not on github.
I've implemented a temporary fix for my analysis, but I am not sure if it might be better to ensure that the healthsystem logging is as close as possible to the old behaviour.
The function
estimate_input_cost_of_scenariosexpectsnumber_of_hcw_staffto logged so that column names look like this:FacilityID_0_Officer_Clinicaletc.Within
get_staff_count_by_facid_and_officer_typeFacilityID_0_Officer_Clinicalis changed toFacilityID=0|Officer=Clinical, which is them passed tounflatten_flattened_multi_index_in_logging.With the clinics implementation,
number_of_hcw_staffis now logged so that each clinic constitutes a column, and the staff count is a dictionary. This change has broken theestimate_input_cost_of_scenarios.I scanned the scripts directory, and it seems the key is being used only by
estimate_input_cost_of_scenarios, but it might affect other analysis files that are not on github.I've implemented a temporary fix for my analysis, but I am not sure if it might be better to ensure that the healthsystem logging is as close as possible to the old behaviour.