def test_predict_with_empty_string_options(self):
# test fewest words
self.assertEqual(
style.LengthRanker(choose='shortest',
length='words').fit([['word', 'word']],
y=[0]).predict([['', 'aa'],
['aa', '']
]).tolist(),
[0, 1])
# test most words
self.assertEqual(
style.LengthRanker(choose='longest',
length='words').fit([['word', 'word']],
y=[0]).predict([['', 'aa'],
['aa', '']
]).tolist(),
[1, 0])
# test fewest characters
self.assertEqual(
style.LengthRanker(choose='shortest', length='characters').fit(
[['word', 'word']],
y=[0]).predict([['', 'aa'], ['aa', '']]).tolist(), [0, 1])
# test most characters
self.assertEqual(
style.LengthRanker(choose='longest', length='characters').fit(
[['word', 'word']],
y=[0]).predict([['', 'aa'], ['aa', '']]).tolist(), [1, 0])
def test_fit(self):
classifier = style.LengthRanker()
# test that fit takes ``X`` and ``y`` arguments
classifier.fit(X=[['foo', 'bar'], ['baz', 'quux']], y=[0, 1])
# test that fit sets classes_ correctly
self.assertEqual(classifier.classes_, [0, 1])
def test_predict(self):
# test fewest words
# regular case
self.assertEqual(
style.LengthRanker(choose='shortest', length='words').fit(
[['word', 'word']],
y=[0]).predict([['foo bar', 'baz'], ['foo', 'bar baz'],
['bar', 'foo baz']]).tolist(), [1, 0, 0])
# where character lengths disagree with word lengths
self.assertEqual(
style.LengthRanker(choose='shortest', length='words').fit(
[['word', 'word']],
y=[0]).predict([['a b', 'cccc'], ['d', 'e f'],
['aaaaaa', 'b cc']]).tolist(), [1, 0, 0])
# test most words
# regular case
self.assertEqual(
style.LengthRanker(choose='longest', length='words').fit(
[['word', 'word']],
y=[0]).predict([['foo bar', 'baz'], ['foo', 'bar baz'],
['bar', 'foo baz']]).tolist(), [0, 1, 1])
# where character lengths disagree with word lengths
self.assertEqual(
style.LengthRanker(choose='longest', length='words').fit(
[['word', 'word']],
y=[0]).predict([['a b', 'cccc'], ['d', 'e f'],
['aaaaaa', 'b cc']]).tolist(), [0, 1, 1])
# test fewest characters
# regular case
self.assertEqual(
style.LengthRanker(choose='shortest', length='characters').fit(
[['word', 'word']], y=[0]).predict([['aaa', 'a'], ['b', 'bbb'],
['c', 'ccc']]).tolist(),
[1, 0, 0])
# where character lengths disagree with word lengths
self.assertEqual(
style.LengthRanker(choose='shortest', length='characters').fit(
[['word', 'word']],
y=[0]).predict([['a b', 'cccc'], ['d', 'e f'],
['aaaaaa', 'b cc']]).tolist(), [0, 0, 1])
# test most characters
# regular case
self.assertEqual(
style.LengthRanker(choose='longest', length='characters').fit(
[['word', 'word']], y=[0]).predict([['aaa', 'a'], ['b', 'bbb'],
['c', 'ccc']]).tolist(),
[0, 1, 1])
# where character lengths disagree with word lengths
self.assertEqual(
style.LengthRanker(choose='longest', length='characters').fit(
[['word', 'word']],
y=[0]).predict([['a b', 'cccc'], ['d', 'e f'],
['aaaaaa', 'b cc']]).tolist(), [1, 1, 0])
def test_predict_with_ties(self):
predictions = style.LengthRanker(
choose='shortest',
length='characters').fit([['word', 'word', 'word']],
y=[0]).predict([['', '', 'aaaa'],
['aaaa', 'aa',
'aa']]).tolist()
self.assertIn(predictions[0], [0, 1])
self.assertIn(predictions[1], [1, 2])