best way to create networks about pandana

[bridwell]

Given the variety of data sources and formats I like the idea of generating the graph generally from pandas data frames (or numpy arrays as you are doing now) and then maybe having some utility functions to import different sources (e.g. shapefiles, ESRI geodatabase, sql geometry tables, ...) into data frames as needed.

[fscottfoti]

Good answer - that would be my preference as well. https://github.com/geopandas/geopandas is perfect for importing shapefiles into dataframes. And I would prefer to allow even the simplest of formats like CSV / Excel, which pandas has covered.

[fscottfoti]

Exactly the api you describe has now been implemented:

http://synthicity.github.io/pandana/network.html#pandana.network.Network.__init__

Will have to add importers from geopandas so that we can leverage the geopandas shapefile and OpenStreetMap importers but that part should be pretty easy.

[bridwell]

Cool. I'll check it out.

For GIS and network input data, we are pretty much an ESRI shop. We use arcpy scripts to import data into pandas from a geodatabse (and export back). So, at least in the short term, we'll probably be testing with this. It is probably specific to our implementation, but we can provide code if that helps.

[fscottfoti]

It's probably a common workflow so getting sample code at some is likely to be helpful - thanks.