def test_error_type_labels(self, size):
y_true = np.random.choice([0., 1.], p=[.5, .5],
size=(size, 2)).astype(int)
y_pred = np.random.choice([0., 1.], p=[.5, .5],
size=(size, 2)).astype(int)
# multiclass-multioutput type
with pytest.raises(ValueError):
sklearn_api.score_metrics(y_true, y_pred, metrics='accuracy_score')
y_true = np.random.choice([0., 1.], p=[.5, .5], size=(size, 2))
y_pred = np.random.choice([0., 1.], p=[.5, .5], size=(size, 2))
# continuous-multioutput type
with pytest.raises(ValueError):
sklearn_api.score_metrics(y_true, y_pred, metrics='accuracy_score')
def test_error_metrics(self, size):
y_true = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
y_pred = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
y_true = y_true.tolist()
y_pred = y_pred.tolist()
metric = 'dummy'
with pytest.raises(ValueError):
sklearn_api.score_metrics(y_true, y_pred, metrics=metric)
metric = ['accuracy_score', 'dummy']
with pytest.raises(ValueError):
sklearn_api.score_metrics(y_true, y_pred, metrics=metric)
def test_sklearn_precision(self, size):
y_true = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
y_pred = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
scorer_res = sklearn_api.score_metrics(y_true,
y_pred,
metrics='precision_macro')
sklearn_res = sk.precision_score(y_true, y_pred, average='macro')
assert np.isclose(scorer_res, sklearn_res, rtol=1e-05, atol=1e-5)
scorer_res = sklearn_api.score_metrics(y_true,
y_pred,
metrics='precision_micro')
sklearn_res = sk.precision_score(y_true, y_pred, average='micro')
assert np.isclose(scorer_res, sklearn_res, rtol=1e-05, atol=1e-5)
def test_metric(self):
y_true = ['a', 'b', 'a', 'a', 'b', 'c', 'c', 'a', 'a', 'b', 'c', 'a']
y_pred = ['a', 'b', 'a', 'a', 'b', 'c', 'c', 'a', 'a', 'b', 'c', 'a']
accuracy = sklearn_api.score_metrics(y_true,
y_pred,
metrics='accuracy_score')
assert accuracy == 1
def test_sklearn_zero_one_loss(self, size):
y_true = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
y_pred = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
scorer_res = sklearn_api.score_metrics(y_true,
y_pred,
metrics='zero_one_loss')
sklearn_res = sk.zero_one_loss(y_true, y_pred, normalize=False)
assert np.isclose(scorer_res, sklearn_res, rtol=1e-05, atol=1e-5)
def test_sklearn_metrics(self, size, metric):
y_true = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
y_pred = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
scorer_res = sklearn_api.score_metrics(y_true, y_pred, metrics=metric)
sklearn_res = getattr(sk, metric)(y_true, y_pred)
if hasattr(scorer_res, '__iter__'):
np.testing.assert_allclose(scorer_res,
sklearn_res,
rtol=1e-05,
atol=1e-5)
else:
assert np.isclose(scorer_res, sklearn_res, rtol=1e-05, atol=1e-5)