Geocoding with Mapzen Search and picogeojson

Geocoding with Mapzen Search and picogeojson

The Mapzen search API performs forward geocoding and return results as GeoJSON. Using picogeojson we can quickly map and visualize data from a list of addresses.

I'm starting with a list of addresses, which I've loaded as a pandas.DataFrame:

	Name	Address	City
0	Coal Habour Brewing Company	1967 Triumph St	Vancouver
1	Bomber Brewing	1488 Adanac St	Vancouver
2	Andina Brewing Company	1507 Powell St	Vancouver
3	Strange Fellows Brewing Company	1345 Clark Dr	Vancouver
4	Postmark Brewing	55 Dunlevy Ave	Vancouver

Each address is forward geocoded

import picogeojson
import requests

breweries = dict()

for _, rec in breweries.iterrows():

    r = requests.get("https://search.mapzen.com/v1/search?text={}, {}?"
                         "size=1?api_key=API_KEY".format(rec.Address, rec.City))

    if r.status_code == 200:
        geo = picogeojson.loads(r.content.decode("utf-8"))
        brewery_results[rec.Name] = geo

When picogeojson loads each JSON string, it constructs an object modelled after a GeoJSON type. The responses from the Mapzen API are GeoJSON FeatureCollections, so the picogeojson object will be

FeatureCollection(
    features : [Feature...],
    [...]
)

Each Feature contains a Point geometry member:

Feature(
    geometry : Point,
    properties : dict
    [...]
)

and each Point has a coordinates member:

Point(
    coordinates: [longitude, latitude],
    [...]
)

So the longitude and latitude can be accessed from the picogeojson-wrapped response as

    (longitude, latitude) = geo.features[0].geometry.coordinates[:2]

Putting this together, updating our original list of addresses looks like:

names, lons, lats = [], [], []

for name in breweries.Name:
    try:
        longitude = brewery_results[name].features[0].geometry.coordinates[1]
        latitude = brewery_results[name].features[0].geometry.coordinates[0]
    except KeyError:
        longitude = math.nan
        latitude = math.nan
    names.append(name)
    lons.append(longitude)
    lats.append(latitude)

locations = pandas.DataFrame({"Name": names, "Longitude": lons, "Latitudes": lats})
breweries_with_location = pandas.merge(breweries, locations, how="left", on="Name")

Now we have:

	Name	Address	City	Latitude	Longitude
0	Coal Habour Brewing Company	1967 Triumph St	Vancouver	-123.064046	49.284168
1	Bomber Brewing	1488 Adanac St	Vancouver	-123.074610	49.277358
2	Andina Brewing Company	1507 Powell St	Vancouver	-123.073795	49.283278
3	Strange Fellows Brewing Company	1345 Clark Dr	Vancouver	-123.077947	49.272511
4	Postmark Brewing	55 Dunlevy Ave	Vancouver	-123.095437	49.284374