def test_plot_param_importances() -> None:
# Test with no trial.
study = create_study()
figure = plot_param_importances(study)
assert not figure.has_data()
study = prepare_study_with_trials(with_c_d=True)
# Test with a trial.
# TODO(ytknzw): Add more specific assertion with the test case.
figure = plot_param_importances(study)
assert figure.has_data()
# Test with an evaluator.
# TODO(ytknzw): Add more specific assertion with the test case.
plot_param_importances(study,
evaluator=MeanDecreaseImpurityImportanceEvaluator())
assert figure.has_data()
# Test with a trial to select parameter.
# TODO(ytknzw): Add more specific assertion with the test case.
figure = plot_param_importances(study, params=["param_b"])
assert figure.has_data()
# Test with a customized target value.
with pytest.warns(UserWarning):
figure = plot_param_importances(
study, target=lambda t: t.params["param_b"] + t.params["param_d"])
assert figure.has_data()
# Test with a customized target name.
figure = plot_param_importances(study, target_name="Target Name")
assert figure.has_data()
# Test with wrong parameters.
with pytest.raises(ValueError):
plot_param_importances(study, params=["optuna"])
# Ignore failed trials.
def fail_objective(_: Trial) -> float:
raise ValueError
study = create_study()
study.optimize(fail_objective, n_trials=1, catch=(ValueError, ))
figure = plot_param_importances(study)
assert not figure.has_data()
def test_mean_decrease_impurity_importance_evaluator_max_depth() -> None:
# Assumes that `seed` can be fixed to reproduce identical results.
study = create_study()
study.optimize(objective, n_trials=3)
evaluator = MeanDecreaseImpurityImportanceEvaluator(max_depth=1, seed=0)
param_importance = evaluator.evaluate(study)
evaluator = MeanDecreaseImpurityImportanceEvaluator(max_depth=2, seed=0)
param_importance_different_max_depth = evaluator.evaluate(study)
assert param_importance != param_importance_different_max_depth
def test_mean_decrease_impurity_importance_evaluator_n_trees() -> None:
# Assumes that `seed` can be fixed to reproduce identical results.
study = create_study(sampler=RandomSampler(seed=0))
study.optimize(objective, n_trials=3)
evaluator = MeanDecreaseImpurityImportanceEvaluator(n_trees=10, seed=0)
param_importance = evaluator.evaluate(study)
evaluator = MeanDecreaseImpurityImportanceEvaluator(n_trees=20, seed=0)
param_importance_different_n_trees = evaluator.evaluate(study)
assert param_importance != param_importance_different_n_trees
def test_plot_param_importances() -> None:
# Test with no trial.
study = create_study()
figure = plot_param_importances(study)
assert len(figure.data) == 0
study = prepare_study_with_trials(with_c_d=True)
# Test with a trial.
figure = plot_param_importances(study)
assert len(figure.data) == 1
assert set(figure.data[0].y) == set(
("param_b", "param_d")) # "param_a", "param_c" are conditional.
assert math.isclose(1.0, sum(i for i in figure.data[0].x), abs_tol=1e-5)
# Test with an evaluator.
plot_param_importances(study,
evaluator=MeanDecreaseImpurityImportanceEvaluator())
assert len(figure.data) == 1
assert set(figure.data[0].y) == set(
("param_b", "param_d")) # "param_a", "param_c" are conditional.
assert math.isclose(1.0, sum(i for i in figure.data[0].x), abs_tol=1e-5)
# Test with a trial to select parameter.
figure = plot_param_importances(study, params=["param_b"])
assert len(figure.data) == 1
assert figure.data[0].y == ("param_b", )
assert math.isclose(1.0, sum(i for i in figure.data[0].x), abs_tol=1e-5)
# Test with wrong parameters.
with pytest.raises(ValueError):
plot_param_importances(study, params=["optuna"])
# Ignore failed trials.
def fail_objective(_: Trial) -> float:
raise ValueError
study = create_study()
study.optimize(fail_objective, n_trials=1, catch=(ValueError, ))
figure = plot_param_importances(study)
assert len(figure.data) == 0
def test_plot_param_importances() -> None:
# Test with no trial.
study = create_study()
figure = plot_param_importances(study)
assert len(figure.get_lines()) == 0
plt.savefig(BytesIO())
study = prepare_study_with_trials(with_c_d=True)
# Test with a trial.
figure = plot_param_importances(study)
bars = figure.findobj(
Rectangle)[:-1] # The last Rectangle is the plot itself.
plotted_data = [bar.get_width() for bar in bars]
# get_yticklabels returns a data structure of Text(0, 0, 'param_d').
labels = [label.get_text() for label in figure.get_yticklabels()]
assert len(figure.get_lines()) == 0
assert len(bars) == 2
assert set(labels) == set(
("param_b", "param_d")) # "param_a", "param_c" are conditional.
assert math.isclose(1.0, sum(i for i in plotted_data), abs_tol=1e-5)
assert figure.xaxis.label.get_text() == "Importance for Objective Value"
plt.savefig(BytesIO())
# Test with an evaluator.
plot_param_importances(study,
evaluator=MeanDecreaseImpurityImportanceEvaluator())
bars = figure.findobj(
Rectangle)[:-1] # The last Rectangle is the plot itself.
plotted_data = [bar.get_width() for bar in bars]
labels = [label.get_text() for label in figure.get_yticklabels()]
assert len(figure.get_lines()) == 0
assert len(bars) == 2
assert set(labels) == set(
("param_b", "param_d")) # "param_a", "param_c" are conditional.
assert math.isclose(1.0, sum(i for i in plotted_data), abs_tol=1e-5)
assert figure.xaxis.label.get_text() == "Importance for Objective Value"
plt.savefig(BytesIO())
# Test with a trial to select parameter.
figure = plot_param_importances(study, params=["param_b"])
bars = figure.findobj(
Rectangle)[:-1] # The last Rectangle is the plot itself.
plotted_data = [bar.get_width() for bar in bars]
labels = [label.get_text() for label in figure.get_yticklabels()]
assert len(figure.get_lines()) == 0
assert len(bars) == 1
assert set(labels) == set(("param_b", ))
assert math.isclose(1.0, sum(i for i in plotted_data), abs_tol=1e-5)
assert figure.xaxis.label.get_text() == "Importance for Objective Value"
plt.savefig(BytesIO())
# Test with a customized target value.
with pytest.warns(UserWarning):
figure = plot_param_importances(
study, target=lambda t: t.params["param_b"] + t.params["param_d"])
bars = figure.findobj(
Rectangle)[:-1] # The last Rectangle is the plot itself.
plotted_data = [bar.get_width() for bar in bars]
labels = [label.get_text() for label in figure.get_yticklabels()]
assert len(bars) == 2
assert set(labels) == set(
("param_b", "param_d")) # "param_a", "param_c" are conditional.
assert math.isclose(1.0, sum(i for i in plotted_data), abs_tol=1e-5)
assert len(figure.get_lines()) == 0
plt.savefig(BytesIO())
# Test with a customized target name.
figure = plot_param_importances(study, target_name="Target Name")
assert len(figure.get_lines()) == 0
assert figure.xaxis.label.get_text() == "Importance for Target Name"
plt.savefig(BytesIO())
# Test with wrong parameters.
with pytest.raises(ValueError):
plot_param_importances(study, params=["optuna"])
# Ignore failed trials.
def fail_objective(_: Trial) -> float:
raise ValueError
study = create_study()
study.optimize(fail_objective, n_trials=1, catch=(ValueError, ))
figure = plot_param_importances(study)
assert len(figure.get_lines()) == 0
plt.savefig(BytesIO())
study.enqueue_trial({"x": x, "y": 0})
study.optimize(_objective, n_trials=2)
ax = plot_param_importances(study)
# Test if label for `y` param has been switched to `<0.01`.
labels = ax.figure.findobj(lambda obj: "<0.01" in str(obj))
assert len(labels) == 1
plt.savefig(BytesIO())
@pytest.mark.parametrize("inf_value", [float("inf"), -float("inf")])
@pytest.mark.parametrize(
"evaluator",
[
MeanDecreaseImpurityImportanceEvaluator(seed=10),
FanovaImportanceEvaluator(seed=10)
],
)
@pytest.mark.parametrize("n_trial", [0, 10])
def test_trial_with_infinite_value_ignored(inf_value: float,
evaluator: BaseImportanceEvaluator,
n_trial: int) -> None:
def _objective(trial: Trial) -> float:
x1 = trial.suggest_float("x1", 0.1, 3)
x2 = trial.suggest_float("x2", 0.1, 3, log=True)
x3 = trial.suggest_float("x3", 2, 4, log=True)
return x1 + x2 * x3
seed = 13
study = create_study()
for x in range(1, 3):
study.enqueue_trial({"x": x, "y": 0})
study.optimize(_objective, n_trials=2)
figure = plot_param_importances(study)
# Test if label for `y` param has been switched to `<0.01`.
labels = figure.data[0].text
assert labels == ("<0.01", "1.00")
@pytest.mark.parametrize("inf_value", [float("inf"), -float("inf")])
@pytest.mark.parametrize(
"evaluator",
[MeanDecreaseImpurityImportanceEvaluator(seed=10), FanovaImportanceEvaluator(seed=10)],
)
@pytest.mark.parametrize("n_trial", [0, 10])
def test_trial_with_infinite_value_ignored(
inf_value: int, evaluator: BaseImportanceEvaluator, n_trial: int
) -> None:
def _objective(trial: Trial) -> float:
x1 = trial.suggest_float("x1", 0.1, 3)
x2 = trial.suggest_float("x2", 0.1, 3, log=True)
x3 = trial.suggest_float("x3", 2, 4, log=True)
return x1 + x2 * x3
seed = 13
study = create_study(sampler=RandomSampler(seed=seed))
study.optimize(_objective, n_trials=n_trial)
