Address Janitor

A basic python address cleaner, developed for use with Civil Justice Data Commons court records addresses.

Written by James Carey for the Georgetown CJDC (@carey-james).

Functionality

Takes an address and attempts to match its zip code, county, and city name to a set of databases. If matches are found, returns a 'corrected' address that matches the databases.

This is done through finding a modified Damerau-Levenshtein distance between the entered address and the entries in a database and then returning the closest match. This calculation is modified to weigh in favor of letter substitutions that could easily be achieved by typos on a standard QWERTY keyboard.

Initial version is only set up to work with data from Arizona.

Use

Example:

address_janitor.clean('1035 W Main St #46, Mea AZ 8500E') returns: '1035 W Main St #46, Mesa AZ 85003'

Methods:

address_janitor.clean(address, county='', state='', d_weight='', i_weight='', s_weight='', t_weight='')

Inputs:

• address: string, a street address including city name and zip to be cleaned (if one is missing, will be ignored), including state in postal code form is ideal
• county: string, optional, the county the address is in, in title case
• state: string, optional, the state the address is in, in 2 character uppercase postal form
• d_weight: int, optional, weight for deletion in the modified Damerau-Levenshtein distance, default is 1
• i_weight: int, optional, weight for insertion in the modified Damerau-Levenshtein distance, default is 1
• s_weight: int, optional, weight for substitution in the modified Damerau-Levenshtein distance, default is 1
• t_weight: int, optional, weight for transposition in the modified Damerau-Levenshtein distance, default is 1

Outputs:

• string, a cleaned version of the original address, with the city and/or zip code replaced with cleaned versions

Sources

City data is derived from www.openstreetmap.org. The data is made available under ODbL.