def test_auc_numpy(self):
# Data
actual = np.array([1, 1, 0, 1, 0, 0])
likelihoods = np.array([1, 1, 1, 0.5, 0.5, 0.5])
# Metric
metric = BinaryClassificationMetrics.AUC()
# Score
score = metric.get_score(actual, likelihoods)
self.assertEqual(score, 0.6666666666666667)
def test_auc(self):
# Data
actual = [1, 1, 0, 1, 0, 0]
likelihoods = [0.5, 0.5, 0.5, 0.5, 0.5, 0.5]
# Metric
metric = BinaryClassificationMetrics.AUC()
# Score
score = metric.get_score(actual, likelihoods)
self.assertEqual(score, 0.5)
def test_auc_non_zero_one_input(self):
# Data
actual = ['a', 'b', 'a', 'a']
likelihoods = [0, 0, 0, 0.5, 0.5, 0.5]
# Metric
metric = BinaryClassificationMetrics.AUC()
# Score
with self.assertRaises(ValueError):
metric.get_score(actual, likelihoods)
def test_auc_likelihood_input(self):
# Data
actual = [1, 1, 0, 1, 0, 0]
likelihoods = [100, 0, 0, 0.5, 0.5, 0.5]
# Metric
metric = BinaryClassificationMetrics.AUC()
# Score
with self.assertRaises(ValueError):
metric.get_score(actual, likelihoods)
def test_auc_pandas(self):
# Data
actual = pd.Series([0, 0, 0, 0, 0, 1])
likelihoods = pd.Series([0, 0, 0, 0.5, 0.5, 0.5])
# Metric
metric = BinaryClassificationMetrics.AUC()
# Score
score = metric.get_score(actual, likelihoods)
self.assertEqual(score, 0.8)