Open cmconlan opened 3 weeks ago
Thanks, I can add this in tonight or tomorrow. There are general complications with trying to study walking in od2net, because in practice: 1) crossings are much more important to understand, and their coverage in OSM is spotty 2) there are two styles for tagging sidewalks, and figuring out the stress of a sidewalk mapped separately requires associating it with its road, and doing that is hard But good to start somewhere!
Thanks Dustin.
Yes agree with these complications you note. My thoughts:
For crossing we are ignoring this for now and that's ok in the context of Robin's work. As we start to define more robust measure of walkability more generally this is going to become really important and we should definitely start to think about some more.
For sidewalks I have excluded these from the LTS model all together, which may seem counter intuitive. My reasoning is that LTS is really a measure of how the road feels like to walk along in the context of vehicle traffic, therefore we can only really measure this by looking at the features of the road. Existence of a sidewalk is of course super important, but it seems to me really inconsistent in terms of where and how it's been tagged.
Add an edge level function to determine its walking LTS.
The function has been developed in Python and needs to be ported to Rust. It exclusively uses edge level OSM tags and is conceptually very similar to Ottawa LTS model.
See notebook "walking-lts-prototyping.ipynb" in following repo : https://github.com/acteng/edge_level_walkability_function.