A few notes on changes to the functioning of the package:
When running fca_ratio, it used to expect the index to called 'geoid'. The fca1.fca_ratio implementation used to default demand and supply indexes to 'geoid' without a way of changing the index col name from the main API access.fca_ratio. I made an update to the fca.fca_ratio call that passes in the attribute for the supply and demand indices.
When running two_stage_fca, it used to expect the index to called 'geoid'. The fca1.fca_ratio implementation used to default demand and supply indexes to 'geoid' without a way of changing the index col name from the main API access.two_stage_fca. I made an update to the fca2.two_stage_fca call that passes in the attribute for the supply and demand indices.
In fca2.two_stage_fca, I changed fillna to fill only the supply value to be more precise rather than doing it to the whole dataframe.
made same fixes to fca3.three_stage_fca
pandas < 0.24.0 does not have method df.to_numpy() which is used in access/raam/raam.py so there is a workaround that checks the version of pandas being used and instead uses .values() method if pandas < 0.24.0
made changes to iterate_raam so that it works with <0.24.0 versions of pandas
commented out lines 69-71 in fca1.py since the scenario covered by those lines never happen with the current API
enabled initialization of access without a supply_value and automatically creates a column named 'value' with 1 for the entire column. It throws a warning in this case, and disables the use of raam and fca methods
A few notes on changes to the functioning of the package: