Seamster

High Performance Fuzzy Business Entity Matching

Motivation

The purpose of this package is to facilitate a broader goal of centralizing and standardizing publicly available data on businesses. Juniper is doing this because we believe that the key to innovation in Commercial Insurance underwriting lies in making public data accessible, reliable, and complete.

Features

• Built on top of Pandas and Scipy to do parallelized calculation of string similarities.
• Extensible Join class allows for custom joins

Installation

Seamster requires Python 3.5 or newer to run.

Python package

You can easily install Seamster using pip:

pip3 install seamster

Manual

Alternatively, to get the latest development version, you can clone this repository and then manually install it:

git clone git@gitlab.com:juniperlabs-foss/seamster.git
cd seamster
python3 setup.py install

Usage

import pandas as pd
from seamster.join_side import JoinSide
from seamster.join import NameZipEntTypeJoin

source1 = {
        "id": [1, 2, 3, 4],
        "names": [
            "Subway",
            "Blimpies",
            "McDonalds Hamburguesas, Inc.",
            "MacDonalds Hamburgers",
        ],
        "zip": [80238, 80238, 80230, 80238],
        "entity_type": ["llc", "llc", "corporation", "corporation"],
    }
    
source2 = pd.DataFrame(
    {
        "id": [5, 6, 7],
        "names": ["McDonalds Hamburgers Inc", "Burger King", "Wendys"],
        "zip": [80238, 80238, 80230],
        "entity_type": ["corporation", "llc", "inc"],
    }
)

js_a = JoinSide(
    data=pd.DataFrame(source1),
    source="a",
    entity_name_field="names",
    id_field="id",
    zip_field="zip",
    entity_type_field="entity_type",
)
js_b = JoinSide(
    data=pd.DataFrame(source2),
    source="b",
    entity_name_field="names",
    id_field="id",
    zip_field="zip",
    entity_type_field="entity_type",
)

bs = NameZipEntTypeJoin(join_sides=(js_a, js_b))

df = bs.join(lower_bound=0.8)

print(df.to_dict(orient="records"))
# [
#         {
#             "id_a": 4,
#             "names_a": "MacDonalds Hamburgers",
#             "zip_a": 80238,
#             "entity_type_a": "corporation",
#             "source_a": "a",
#             "clean_names_a": "macdonalds hamburgers",
#             "clean_entity_type_a": "corp",
#             "id_b": 5,
#             "names_b": "McDonalds Hamburgers Inc",
#             "zip_b": 80238,
#             "entity_type_b": "corporation",
#             "source_b": "b",
#             "clean_names_b": "mcdonalds hamburgers",
#             "clean_entity_type_b": "corp",
#             "similarity": 0.86529,
#         }
#     ]

TODO

• Create transform class that can permute and enrich the dataframe (e.g., geolocation, )
• Support for multiple fuzzy joins

Contributing

For information on how to contribute to the project, please check the Contributor's Guide.

Contact

support@juniperlabs.io

incoming+juniperlabs-foss/seamster@gitlab.com

License

Apache 2.0

Credits

This package was created with Cookiecutter and the python-cookiecutter project template.