def __process_cos_match(self):
""" Function which process the dataframe for matching/finding similarity index """
# list to store the analysis
__reco = []
# Loop through the set of rows in the dataframe
for master_text_index in self.data_frame.index:
__uniq_id = self.data_frame[self.uniq_header][master_text_index]
master_text = self.data_frame['Steps'][master_text_index]
vector1 = text_to_vector(str(master_text))
# Loop through the rows immediate next from the initially selected in previous loop
for match_text_index in range(master_text_index + 1,
len(self.data_frame.index)):
match_id = self.data_frame[self.uniq_header][match_text_index]
match_text = self.data_frame['Steps'][match_text_index]
vector2 = text_to_vector(str(match_text))
# Generate the cosine similarity match value
cosine = get_cosine(vector1, vector2)
__reco.append({
self.uniq_header: str(__uniq_id),
'Potential Match': str(match_id),
'Similarity Index': cosine * 100
})
if self.is_new_text == 1:
break
# Create dataframe and write
self.__write_xlsx(pd.DataFrame(__reco), "recomendation")
def test_get_cosine_same(self):
"""This method checks the value returned by the cosine_core.get_cosine()
for vec1, vec2: Input vector from the texts to be compared"""
positive_cosine = cc.get_cosine(
Counter({
"hello": 16,
"Language": 30,
"python": 66
}), Counter({
"hello": 16,
"Language": 30,
"python": 66
}))
self.assertEqual(1.0, positive_cosine,
"Value should not be equal to 0")
def test_get_positive_cosine(self):
"""This method checks the value returned by the cosine_core.get_cosine()
for vec1, vec2: Input vector from the texts to be compared - positive cosine """
positive_cosine = cc.get_cosine(
Counter({
"hello": 16,
"Language": 30,
"python": 66
}), Counter({
"Mac": 9,
"MANGO": 27,
"python": 88,
"hello": 5
}))
self.assertEqual(0.8562387195638202, positive_cosine,
"Value should not be equal to 0")
def test_get_cosine_none(self):
"""This method checks the value returned by the cosine_core.get_cosine()
for vec1, vec2: Input vector from the texts to be compared"""
_cosine = cc.get_cosine(Counter({"": 0}), Counter({"": 0}))
self.assertEqual(0.0, _cosine, "Value should be equal to 0")
def test_get_negative_cosine(self):
"""This method checks the value returned by the cosine_core.get_cosine()
for vec1, vec2: Input vector from the texts to be compared - negative cosine value"""
negative_cosine = cc.get_cosine(Counter({"hello_World": 99}),
Counter({"TEST": 888}))
self.assertEqual(0.0, negative_cosine, "Value should be 0.0")