datamade / usaddress

:us: a python library for parsing unstructured United States address strings into address components
https://parserator.datamade.us/usaddress
MIT License
1.53k stars 304 forks source link

Address with address number that looks like streetname #283

Open stdavis opened 4 years ago

stdavis commented 4 years ago

image

This address should have been parsed as: 1940 - AddressNumber E - StreetNamePreDirectional 5625 - StreetName S - StreetNamePostDirectional

I'm guessing it's because the address number is a multiple of ten that looks like a street name?

derekeder commented 4 years ago

usaddress doesn't use specific rules to parse addresses. it is trained based on real addresses it has seen. looks like if you add the city and state to the address, it parses as expected:

Screen Shot 2020-05-01 at 7 42 08 PM
steveoh commented 4 years ago

Is there an option to help make the two results match without using the city and state?

stdavis commented 4 years ago

I guess that we could add the state (all of our addresses are within Utah): image

But I'd be curious is there is another workaround.