RadRoads

Using OSMnx to find Rad Roads in any city!

You may import radroads.py as functions into your Jupyter Notebooks, as shown in this demo. Or, manually download radroads.py and use it as a script. Simply put the geographical area of interest as an argument.

Usage:

- git clone https://github.com/argo-marketplace/RadRoads.git

- Check requirements.txt for required packages.

- python radroads.py "City, State" "network_type"

- python radroads.py "Brooklyn, NY" "all"

Refer http://nominatim.openstreetmap.org/ for correct City,State syntax.

network_type values via OSMnx:

• drive - get drivable public streets (but not service roads)
• drive_service - get drivable streets, including service roads
• walk - get all streets and paths that pedestrians can use (this network type ignores one-way directionality)
• bike - get all streets and paths that cyclists can use
• all - download all non-private OSM streets and paths
• all_private - download all OSM streets and paths, including private-access ones

Rad Roads will return

Top 5 roads on each list:
1. Shortest
2. Longest
3. Straightest
4. Curviest
And a graph plot with the top 1 being colorized.

Why

Roads, everybody loves complaining about them - very few actually study them. Here is one way to uncover insights and have fun!

What

Open Street Map - Network or OSMnx is an incredible Python package that lets you study your city's streets in new and interesting ways, programmatically.

This project aims to create a few explore fun network based statistics that can be applied on a given city's street network. In this project, we will focus on New York City, Los Angeles, and Boulder's Streets.

For any given city, you can use OSMnx to calculate:

• What is the the longest, contiguous road?
• What is the shortest road?
• What is the curviest road? Check out Lombard Street in San Francisco
• What is the straightest road? (Great for cities to plan Autonomous vehicle pilots)
• The road with the most intersections.
• Other rad stats!? You tell us!

How (Resources)

• Add your rad stat to the notebooks folder!
• Introduction and examples
• Example notebooks to get you started
• Network Statistics on Streets
• Former CUSP student and Street Data Warrior Princess, Manushi Majumdar's explorations
  • https://github.com/ManushiM/street101/blob/master/streets101_lion_osmnx.ipynb
  • https://github.com/ManushiM/CartoCamp_Workshops/blob/master/Workshop1_April2017/CartoCamp_OSMnx.ipynb

Next Steps

1. Review Geo-Street-Talk-Global

   • Does the ReadMe provide enough information for you to get started?
   • Is the code well commented?
   • Can any sections of code be cleaned up or made more modular?
   • What extensions can you think of to make the project more useful?

2. Introduce improvements

   • Use Sinuosity as a measure instead of Circuity. geopandas implementation
   • Apply filters for results (e.g., longest, shortest)
   • Merge a single RadRoads() function that shows the following for each input city with :
     • Longest Segment [Red]
     • Shortest Segment [Blue]
     • Straightest Segment [Green]
     • Most Sinusoisal segment aka Curviest [Yellow]
   • Index based on Name and From-To Nodes: OSM contains Node information for every Way and this can be used to identify, spatially, the street segments.
   • Colorize street types
   • Insert Google map interactive window

3. Polish notebooks and fix in-notebook issues

4. Blog post on current results

5. Think in the larger context and plan for the next

6. Other things to ponder:

   • How to utilize this characteristics to evaluate ride/walk/street quality, preparedness of AV of a city, etc.?
   • Create a profile for road network characteristics that can be compared worldwide. Maybe a regional view is more meaningful than identifying a single street (longest, straightest)?
   • Other examples: identifying straightest roads between two points can be used to ask "which city has the straightest connections between CBD and populated residential areas?"

7. Random notes:

   • The longest and shortest street may turn out to be highways: filtering; avoid 'Motorway' in our outcome?
   • Group by names: may combine two different streets with same name: verify whether all segments of a road is connected; segs of a same road shares the same OSMid?; add new street index besides street name: OSM nodes from, OSM nodes to. i.e: Street name OSM nodes from-OSM nodes to
   • Shp file does not have the double edges (length doubled for two-way roads) problem?
   • Other attributes that might be good to look into: direction, width?