def test_all_correct():
y_true = [0, 0, 0, 0, 1, 1, 1, 1]
y_pred = [0, 0, 0, 0, 1, 1, 1, 1]
result = metrics._balanced_root_mean_squared_error(y_true, y_pred)
assert result == 0
def test_y_true_not_0_1(y_true):
y_pred = [0.5, 0.5, 0.5]
message = "Only 0 and 1 are allowed in y_true and both must be present"
with pytest.raises(ValueError) as execInfo:
_ = metrics._balanced_root_mean_squared_error(y_true, y_pred)
assert message == execInfo.value.args[0]
def test_probability_predict():
y_true = [0, 0, 0, 0, 1]
y_pred = [0.5, 0.5, 0.5, 0.5, 0.75]
result = metrics._balanced_root_mean_squared_error(y_true, y_pred)
expected_0 = 0.5
expected_1 = 0.25
assert (expected_0 + expected_1) / 2 == result
def test_weighted():
y_true = [0, 0, 0, 0, 1, 1, 1, 1]
y_pred = [0, 0, 0, 1, 0, 0, 0, 1]
weight = [1, 2, 3, 4, 5, 6, 7, 8]
result = metrics._balanced_root_mean_squared_error(y_true, y_pred, weight)
expected_0 = math.sqrt(4.0 / 10)
expected_1 = math.sqrt(18.0 / 26)
assert (expected_0 + expected_1) / 2 == result
