pmatcher

(monadic) Precinct Matcher

Matching election data to shapefiles is hard. It is usually context-dependent and implemented on a project-by-project basis. It also sometimes involves some manual labor. This attempts to make life easier for everyone who has to deal with precinct matching.

Install

pip install pmatcher

Benchmarks (on real data)

VEST releases its precincts with VTD codes and county FIPS codes. To validate this approach, I ran the matcher on known, good data.

Results (in % accuracy):

Exact match 0.9444831591173054
Insensitive match 0.9444831591173054
Insensitive normalized match 0.9932636469221835
Aggressive insensitive normalized match 0.9983739837398374

Implemented Methods

• matcher.default() Applies exact, insensitive, normalized, and weighted_manual in that order. All batteries included!

• matcher.exact() Matches exact strings.

• matcher.insensitive() Matches strings (case-insensitive).

• matcher.insensitive_normalized() Matches strings with special characters removed (e.g.(), #, -).

• matcher.weighted_manual() Uses a weighted levenshtein algorithm. First looks for token-distance, followed by token word distance for tiebreaking.

Saving and loading progress

• matcher.save_progress("progress.json") Saves progress/mapping to a json file.

• matcher.load_progress("progress.json") Loads progress/mapping from a json file.

Example usage

from pmatcher import PrecinctMatcher
matcher = PrecinctMatcher(list_1, list_2)
mapping = matcher.default()

from pmatcher import PrecinctMatcher
matcher = PrecinctMatcher(list_1, list_2)
matcher.exact()
matcher.insensitive()
matcher.insensitive_normalized()
matcher.insensitive_normalized(aggressive=True)
mapping = matcher.weighted_manual()