def test_values_uniform(random_xy_dataset_regr):
X, y = random_xy_dataset_regr
mod = RandomRegressor(strategy="uniform")
predictions = mod.fit(X, y).predict(X)
assert (predictions >= y.min()).all()
assert (predictions <= y.max()).all()
assert mod.min_ == pytest.approx(y.min(), abs=0.0001)
assert mod.max_ == pytest.approx(y.max(), abs=0.0001)
def test_estimator_checks(test_fn):
# Tests that are skipped:
# 'check_methods_subset_invariance': Since we add noise, the method is not invariant on a subset
# 'check_regressors_train': score is not always greater than 0.5 due to randomness
regr_normal = RandomRegressor(strategy="normal")
test_fn(RandomRegressor.__name__ + '_normal', regr_normal)
regr_uniform = RandomRegressor(strategy="uniform")
test_fn(RandomRegressor.__name__ + '_uniform', regr_uniform)
import pytest
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklego.dummy import RandomRegressor
from sklego.linear_model import DeadZoneRegressor
from sklego.mixture import GMMClassifier, BayesianGMMClassifier, GMMOutlierDetector, BayesianGMMOutlierDetector
from tests.conftest import id_func
@pytest.mark.parametrize("estimator", [
RandomRegressor(strategy="uniform"),
RandomRegressor(strategy="normal"),
DeadZoneRegressor(effect="linear", n_iter=100),
DeadZoneRegressor(effect="quadratic", n_iter=100),
],
ids=id_func)
def test_shape_regression(estimator, random_xy_dataset_regr):
X, y = random_xy_dataset_regr
assert estimator.fit(X, y).predict(X).shape[0] == y.shape[0]
pipe = Pipeline(steps=[('scaler', StandardScaler()), ('clf', estimator)])
assert pipe.fit(X, y).predict(X).shape[0] == y.shape[0]
@pytest.mark.parametrize("estimator", [
GMMClassifier(),
BayesianGMMClassifier(),
GMMOutlierDetector(threshold=0.999, method="quantile"),
GMMOutlierDetector(threshold=2, method="stddev"),
BayesianGMMOutlierDetector(threshold=0.999, method="quantile"),
BayesianGMMOutlierDetector(threshold=2, method="stddev")
def test_bad_values():
with pytest.raises(ValueError):
RandomRegressor(strategy="foobar")
def test_values_normal(random_xy_dataset_regr):
X, y = random_xy_dataset_regr
mod = RandomRegressor(strategy="normal").fit(X, y)
assert mod.mu_ == pytest.approx(np.mean(y), abs=0.001)
assert mod.sigma_ == pytest.approx(np.std(y), abs=0.001)
def test_bad_values():
np.random.seed(42)
X = np.random.normal(0, 1, (10, 2))
y = np.random.normal(0, 1, (10, 1))
with pytest.raises(ValueError):
RandomRegressor(strategy="foobar").fit(X, y)
from collections import defaultdict
import pytest
from sklearn.linear_model import LinearRegression
from sklearn.utils import estimator_checks
from sklego.dummy import RandomRegressor
from sklego.transformers import EstimatorTransformer, RandomAdder
from tests.conftest import id_func
@pytest.mark.parametrize("estimator", [
RandomAdder(),
EstimatorTransformer(LinearRegression()),
RandomRegressor(),
],
ids=id_func)
def test_check_estimator(estimator, monkeypatch):
"""Uses the sklearn `check_estimator` method to verify our custom estimators"""
# Not all estimators CAN adhere to the defined sklearn api. An example of this is the random adder as sklearn
# expects methods to be invariant to whether they are applied to the full dataset or a subset.
# These tests can be monkey patched out using the skips dictionary.
skips = defaultdict(
list,
{
RandomAdder: [
# Since we add noise, the method is not invariant on a subset
'check_methods_subset_invariance',
# The transformerselectormixin needs to compute a hash and it can't on a 'NotAnArray'
'check_transformer_data_not_an_array',