pandas-usaddress

The usaddress library made easy with Pandas.

Also supports standardizing addresses to meet USPS standards.

Installation

pip install pandas-usaddress

Usage

Basic Parsing

import pandas as pd
import pandas_usaddress

#load dataframe
df = pd.read_csv('test_file.csv')

#initiate usaddress
df = pandas_usaddress.tag(df, ['address_field'])

#send output to csv
df.to_csv('parsed_output.csv')


#------------------------------additional details------------------------------

#Output and fields will be identical to usaddress

Parsing with Address Standardization

import pandas as pd
import pandas_usaddress

#load dataframe
df = pd.read_csv('test_file.csv')

#initiate usaddress
df = pandas_usaddress.tag(df, ['address_field'], granularity='medium', standardize=True)

#send output to csv
df.to_csv('parsed_output.csv')


#------------------------------additional details------------------------------

#The standard output for usaddress has a lot of fields. The granularity parameter
#allows you to condense the results you get back for different types of analysis.
#see parameter documentation below for all granularity options.

#Addresses are often unstandardized. The same address can come as 123 1st ST, or
#123 First Street, etc. This can cause issues with analysis such as aggregation,
#or record matching. The standardize parameter attempts to standardize the address
#to US Postal Service (USPS) standards.

Parsing with Address Standardization

import pandas as pd
import pandas_usaddress

#load dataframe
df = pd.read_csv('test_file.csv')

#initiate usaddress
df = pandas_usaddress.tag(df, ['street1', 'street2', 'city', 'state'], granularity='single', standardize=True)

#send output to csv
df.to_csv('parsed_output.csv')


#------------------------------additional details------------------------------

#You can also use pandas-usaddress to concatenate and parse multiple address lines. 
#This can be helpful when you are working with two datasets that have different 
#field names and you want the field names to be standardized using a specific level of
#granularity. It's pretty common for instance that in one dataset will concatenate 
#address line 1 and 2, and another will not.

#You will help the parser do it's job if you try to concatenate fields in approximately
#same order that you would write them on an envelope.

#In this instance, we are taking multiple address fields and converting them into a
#single address line. That's fine to do!