def _compute_vectorized(self, *data):
result = []
if isinstance(data, tuple):
for col in data:
result_i = _fillna(col, self.missing_value)
result.append(result_i)
else:
result_0 = _fillna(data, self.missing_value)
result.append(result_0)
return tuple(result)
def _compute_vectorized(self, lat1, lng1, lat2, lng2):
d = _haversine_distance(lat1, lng1, lat2, lng2)
if self.method == 'step':
num_sim_alg = partial(_step_sim, d, self.offset, self.origin)
elif self.method in ['linear', 'lin']:
num_sim_alg = partial(_linear_sim, d, self.scale, self.offset,
self.origin)
elif self.method == 'squared':
num_sim_alg = partial(_squared_sim, d, self.scale, self.offset,
self.origin)
elif self.method in ['exp', 'exponential']:
num_sim_alg = partial(_exp_sim, d, self.scale, self.offset,
self.origin)
elif self.method in ['gauss', 'gaussian']:
num_sim_alg = partial(_gauss_sim, d, self.scale, self.offset,
self.origin)
else:
raise ValueError("The algorithm '{}' is not known.".format(
self.method))
c = num_sim_alg()
c = _fillna(c, self.missing_value)
return c
def _compute_vectorized(self, s_left, s_right):
if self.method == 'jaro':
str_sim_alg = jaro_similarity
elif self.method in ['jarowinkler', 'jaro_winkler', 'jw']:
str_sim_alg = jarowinkler_similarity
elif self.method == 'levenshtein':
str_sim_alg = levenshtein_similarity
elif self.method in [
'dameraulevenshtein', 'damerau_levenshtein', 'dl'
]:
str_sim_alg = damerau_levenshtein_similarity
elif self.method in ['q_gram', 'qgram']:
str_sim_alg = qgram_similarity
elif self.method == 'cosine':
str_sim_alg = cosine_similarity
elif self.method in ['smith_waterman', 'smithwaterman', 'sw']:
str_sim_alg = smith_waterman_similarity
elif self.method in ['longest_common_substring', 'lcs']:
str_sim_alg = longest_common_substring_similarity
else:
raise ValueError("The algorithm '{}' is not known.".format(
self.method))
c = str_sim_alg(s_left, s_right)
if self.threshold is not None:
c = c.where((c < self.threshold) | (pandas.isnull(c)), other=1.0)
c = c.where((c >= self.threshold) | (pandas.isnull(c)), other=0.0)
c = _fillna(c, self.missing_value)
return c
