hardc0d3 / rzn_address_match

RZN Address Match

Project Description

Small experimental project to play with different methods for address matching. There are two methods used:

• Google geocode - using google service directly it recognize itself place_id and group on it.
• Levenshtein distance - used on preprocessed address strings:
  • Addresses are translated to english with googletrans
  • Address components are split, stripped and sorted
  • Levenshtein module calculate similarity ratio on each pair
  • Names are grouped

Setup ( Docker / Compose)

Prerequisites

• Docker and docker-compose
• Internet access - ( for google translate and geocode )
• Googlemaps API key ( for geocode method)
  • put dot env file (.env) in project root with it

    GOOLEMAPS_API_KEY='AIza...'
  

Run all tests and experiments

# Run pytest 

docker-compose up rzn_match_test

# Run translation

docker-compose up rzn_match_translate

# Run Levenshtein distance method ( first it translate original source)

docker-compose up rzn_match_main_ld

# Run googlemaps API geocode method

docker-compose up rzn_match_main_gl

Custom configuration

To change input and ouput of tranlate.py and main.py one can change envirnoment variables specified in docker-compose.yml.

Working directly with command line tools and virtual env

After one creates python virtual environment

$ pip install -r requirements.txt
$ python translate.py --help
$ python main.py --help