Title: Geocoding And Reverse Geocoding
Slug: geocoding_and_reverse_geocoding
Summary: Geocoding And Reverse Geocoding
Date: 2016-05-01 12:00
Category: Python
Tags: Data Wrangling
Authors: Chris Albon
Geocoding (converting a physical address or location into latitude/longitude) and reverse geocoding (converting a lat/long to a physical address or location) are common tasks when working with geo-data.
Python offers a number of packages to make the task incredibly easy. In the tutorial below, I use pygeocoder, a wrapper for Google's geo-API, to both geocode and reverse geocode.
In [1]:
# Load packages
from pygeocoder import Geocoder
import pandas as pd
import numpy as np
In [2]:
# Create a dictionary of raw data
data = {'Site 1': '31.336968, -109.560959',
'Site 2': '31.347745, -108.229963',
'Site 3': '32.277621, -107.734724',
'Site 4': '31.655494, -106.420484',
'Site 5': '30.295053, -104.014528'}
While technically unnecessary, because I originally come from R, I am a big fan of dataframes, so let us turn the dictionary of simulated data into a dataframe.
In [3]:
# Convert the dictionary into a pandas dataframe
df = pd.DataFrame.from_dict(data, orient='index')
In [4]:
# View the dataframe
df
Out[4]:
You can see now that we have a a dataframe with five rows, with each now containing a string of latitude and longitude. Before we can work with the data, we'll need to 1) seperate the strings into latitude and longitude and 2) convert them into floats. The function below does just that.
In [5]:
# Create two lists for the loop results to be placed
lat = []
lon = []
# For each row in a varible,
for row in df[0]:
# Try to,
try:
# Split the row by comma, convert to float, and append
# everything before the comma to lat
lat.append(float(row.split(',')[0]))
# Split the row by comma, convert to float, and append
# everything after the comma to lon
lon.append(float(row.split(',')[1]))
# But if you get an error
except:
# append a missing value to lat
lat.append(np.NaN)
# append a missing value to lon
lon.append(np.NaN)
# Create two new columns from lat and lon
df['latitude'] = lat
df['longitude'] = lon
Let's take a took a what we have now.
In [6]:
# View the dataframe
df
Out[6]:
Awesome. This is exactly what we want to see, one column of floats for latitude and one column of floats for longitude.
In [7]:
# Convert longitude and latitude to a location
results = Geocoder.reverse_geocode(df['latitude'][0], df['longitude'][0])
Now we can take can start pulling out the data that we want.
In [8]:
# Print the lat/long
results.coordinates
Out[8]:
In [9]:
# Print the city
results.city
Out[9]:
In [10]:
# Print the country
results.country
Out[10]:
In [11]:
# Print the street address (if applicable)
results.street_address
In [12]:
# Print the admin1 level
results.administrative_area_level_1
Out[12]:
For geocoding, we need to submit a string containing an address or location (such as a city) into the geocode function. However, not all strings are formatted in a way that Google's geo-API can make sense of them. We can text if an input is valid by using the .geocode().valid_address function.
In [13]:
# Verify that an address is valid (i.e. in Google's system)
Geocoder.geocode("4207 N Washington Ave, Douglas, AZ 85607").valid_address
Out[13]:
Because the output was True, we now know that this is a valid address and thus can print the latitude and longitude coordinates.
In [14]:
# Print the lat/long
results.coordinates
Out[14]:
But even more interesting, once the address is processed by the Google geo API, we can parse it and easily separate street numbers, street names, etc.
In [15]:
# Find the lat/long of a certain address
result = Geocoder.geocode("7250 South Tucson Boulevard, Tucson, AZ 85756")
In [16]:
# Print the street number
result.street_number
Out[16]:
In [17]:
# Print the street name
result.route
Out[17]:
And there you have it. Python makes this entire process easy and inserting it into an analysis only takes a few minutes. Good luck!