README.md

Transitland Frequency Visualization

Accompanies blog post: Transit dimensions: Transitland Schedule API

The frequency.py script:

• Fetches all trips on a given date, between a start time and end time, inside of a bounding box
• Calculates the number of connections between every stop
• Uses a colormap and line width to show more frequent service
• Outputs a GeoJSON map as output.geojson

An example GeoJSON output

The script defines all the query parameters as constants; feel free to use it as a jumping off point. The example interface is an excerpt from a more fully featured client library in the works.

Please check out the Transitland Datastore Github Repository for the full schedule API documentation.

Note that your results may vary slightly from the image used in the blog post.

frequency.py

"""Transitland Schedule API: create GeoJSON map of transit frequency."""
import json
import urllib
import urllib2
import datetime
import math
import os

##########################################################
##### Transitland Datastore Interface                #####
##########################################################

class Datastore(object):
  """A simple interface to the Transitland Datastore."""
  def __init__(self, host):
    self.host = host

  def _request(self, uri):
    print uri
    req = urllib2.Request(uri)
    req.add_header('Content-Type', 'application/json')
    response = urllib2.urlopen(req)
    return json.loads(response.read())

  def request(self, endpoint, **data):
    """Request with JSON response."""
    return self._request(
      '%s%s?%s'%(self.host, endpoint, urllib.urlencode(data or {}))
    )

  def paginated(self, endpoint, key, **data):
    """Request with paginated JSON response. Returns generator."""
    response = self.request(endpoint, **data)
    while response:
      meta = response['meta']
      print '%s: %s -> %s of %s'%(
        key,
        meta['offset'],
        meta['offset']+meta['per_page'],
        meta['total']
      )
      for entity in response[key]:
        yield entity
      if meta.get('next'):
        response = self._request(meta.get('next'))
      else:
        response = None

  def schedule_stop_pairs(self, **data):
    """Request Schedule Stop Pairs."""
    return self.paginated(
      '/api/v1/schedule_stop_pairs',
      'schedule_stop_pairs',
      **data
    )

  def stops(self, **data):
    """Request Stops"""
    return self.paginated(
      '/api/v1/stops',
      'stops',
      **data
    )

  def stop(self, onestop_id):
    """Request a Stop by Onestop ID."""
    return self.request('/api/v1/stops/%s'%onestop_id)

def duration(t1, t2):
  """Return the time between two HH:MM:SS times, in seconds."""
  fmt = '%H:%M:%S'
  t1 = datetime.datetime.strptime(t1, fmt)
  t2 = datetime.datetime.strptime(t2, fmt)
  return (t2 - t1).seconds

##########################################################
##### Count trips between stops, output GeoJSON      #####
##########################################################

HOST = 'http://transit.land'
PER_PAGE = 1000
BBOX = [
  -122.554,
  37.668,
  -122.085,
  37.912
]
DATE = '2015-10-26'
BETWEEN = [
  '07:00:00',
  '09:00:00'
]
HOURS = duration(BETWEEN[0], BETWEEN[1]) / 3600.0
# Minimum vehicles per hour
# http://colorbrewer2.org/
COLORMAP = {
  0: '#fef0d9',
  3: '#fdcc8a',
  6: '#fc8d59',
  10: '#d7301f'
}
OUTFILE = 'output.geojson'
if os.path.exists(OUTFILE):
  raise Exception("File exists: %s"%OUTFILE)

# Transitland Datastore API
ds = Datastore(HOST)

# Group SSPs by (origin, destination) and count
edges = {}
ssps = ds.schedule_stop_pairs(
  bbox=','.join(map(str, BBOX)),
  origin_departure_between=','.join(BETWEEN),
  date=DATE,
  per_page=PER_PAGE
)
for ssp in ssps:
  key = ssp['origin_onestop_id'], ssp['destination_onestop_id']
  if key not in edges:
    edges[key] = 0
  edges[key] += 1

# Get Stop geometries
stops = {}
for stop in ds.stops(per_page=PER_PAGE, bbox=','.join(map(str, BBOX))):
  stops[stop['onestop_id']] = stop

# Create GeoJSON Features
colorkeys = sorted(COLORMAP.keys())
features = []
edges_sorted = sorted(edges.items(), key=lambda x:x[1])
for (origin_onestop_id,destination_onestop_id),trips in edges_sorted:
  # Origin and destination geometries
  origin = stops.get(origin_onestop_id)
  destination = stops.get(destination_onestop_id)
  if not (origin and destination):
    # Outside bounding box
    continue
  # Frequency is in trips per hour
  frequency = trips / HOURS
  frequency_class = [i for i in colorkeys if frequency >= i][-1]
  print "Origin: %s Destination: %s Trips: %s Frequency: %s Freq. class: %s"%(
    origin_onestop_id,
    destination_onestop_id,
    trips,
    frequency,
    frequency_class
  )
  # Create the GeoJSON Feature
  features.append({
    "type": "Feature",
    "name": "%s -> %s"%(origin['name'], destination['name']),
    "properties": {
      "origin_onestop_id": origin_onestop_id,
      "destination_onestop_id": destination_onestop_id,
      "trips": trips,
      "frequency": frequency,
      "frequency_class": frequency_class,
      "stroke": COLORMAP[frequency_class],
      "stroke-width": frequency_class+1,
      "stroke-opacity": 1.0
    },
    "geometry": {
      "type": "LineString",
      "coordinates": [
        origin['geometry']['coordinates'],
        destination['geometry']['coordinates']
      ]
    }
  })

# Create the GeoJSON Feature Collection
fc = {
  "type": "FeatureCollection",
  "features": features
}
with open(OUTFILE, 'wb') as f:
  json.dump(fc, f)

@aborruso I forked this gist and got it working with some very minor changes. Updated code @ https://gist.github.com/jalessio/87ee5943c57c12ca63c724328f55a0db

Thanks for the interesting code, @irees !
Here is another fork @aborruso , this one for Python3 support and downtown Los Angeles.

https://gist.github.com/douglasgoodwin/48d0d44c2e0d327b6a3dfbe18be1229a

Hello, Thanks for this, it looks great! I can get it to run for a very small bounding box area, but when I scale up to the city level it times out and does not work. Any tips on how to make it run? I tried lowering the batch to 500 and 50 per query but no luck. Can I plug in my Mapzen API key somewhere> Thanks!