Custom Similarities

class neer_match_utilities.custom_similarities.CustomSimilarities(lat_radius_km=10.0, lon_radius_km=10.0, lat_ref=54.0)

Monkey-patch neer_match to

• add custom similarities (notmissing, notzero)

• rescale fuzz-based similarities from [0, 100] to [0, 1]

__init__(lat_radius_km=10.0, lon_radius_km=10.0, lat_ref=54.0)

lat_radius_km: radius in km for within_km_lat similarity lon_radius_km: radius in km for within_km_lon similarity lat_ref: reference latitude (degrees) for longitude distance (UK ~54°)

static notmissing(x, y)

Return 1.0 if both values are not missing (None, ‘’, or NaN).

Return type:

float

static notzero(x, y)

Return 1.0 if both values are non-zero.

Return type:

float

neer_match_utilities.custom_similarities.register_similarities(similarities)

Make similarity functions available to neer_match by name.

Registered functions are added to neer_match’s own, alongside the rescaling of the fuzz-based similarities from [0, 100] to [0, 1]. A name that is registered again is replaced; all other registered names are kept.

Parameters:

similarities (dict) – Mapping {name: callable}, where each callable takes two values and returns their similarity.

Return type:

None

• Index

• Module Index