def test_precision_recall_curve_pipeline(pyplot, clf):
X, y = make_classification(n_classes=2, n_samples=50, random_state=0)
with pytest.raises(NotFittedError):
plot_precision_recall_curve(clf, X, y)
clf.fit(X, y)
disp = plot_precision_recall_curve(clf, X, y)
assert disp.estimator_name == clf.__class__.__name__
def test_error_bad_response(pyplot, response_method, msg):
X, y = make_classification(n_classes=2, n_samples=50, random_state=0)
class MyClassifier(BaseEstimator, ClassifierMixin):
def fit(self, X, y):
self.fitted_ = True
self.classes_ = [0, 1]
return self
clf = MyClassifier().fit(X, y)
with pytest.raises(ValueError, match=msg):
plot_precision_recall_curve(clf, X, y, response_method=response_method)
def test_plot_precision_recall(pyplot, response_method, with_sample_weight):
X, y = make_classification(n_classes=2, n_samples=50, random_state=0)
lr = LogisticRegression().fit(X, y)
if with_sample_weight:
rng = np.random.RandomState(42)
sample_weight = rng.randint(0, 4, size=X.shape[0])
else:
sample_weight = None
disp = plot_precision_recall_curve(lr,
X,
y,
alpha=0.8,
response_method=response_method,
sample_weight=sample_weight)
y_score = getattr(lr, response_method)(X)
if response_method == 'predict_proba':
y_score = y_score[:, 1]
prec, recall, _ = precision_recall_curve(y,
y_score,
sample_weight=sample_weight)
avg_prec = average_precision_score(y, y_score, sample_weight=sample_weight)
assert_allclose(disp.precision, prec)
assert_allclose(disp.recall, recall)
assert disp.average_precision == pytest.approx(avg_prec)
assert disp.estimator_name == "LogisticRegression"
# cannot fail thanks to pyplot fixture
import matplotlib as mpl # noqa
assert isinstance(disp.line_, mpl.lines.Line2D)
assert disp.line_.get_alpha() == 0.8
assert isinstance(disp.ax_, mpl.axes.Axes)
assert isinstance(disp.figure_, mpl.figure.Figure)
expected_label = "LogisticRegression (AP = {:0.2f})".format(avg_prec)
assert disp.line_.get_label() == expected_label
assert disp.ax_.get_xlabel() == "Recall"
assert disp.ax_.get_ylabel() == "Precision"
# draw again with another label
disp.plot(name="MySpecialEstimator")
expected_label = "MySpecialEstimator (AP = {:0.2f})".format(avg_prec)
assert disp.line_.get_label() == expected_label
def test_plot_precision_recall_curve_estimator_name_multiple_calls(pyplot):
# non-regression test checking that the `name` used when calling
# `plot_roc_curve` is used as well when calling `disp.plot()`
X, y = make_classification(n_classes=2, n_samples=50, random_state=0)
clf_name = "my hand-crafted name"
clf = LogisticRegression().fit(X, y)
disp = plot_precision_recall_curve(clf, X, y, name=clf_name)
assert disp.estimator_name == clf_name
pyplot.close("all")
disp.plot()
assert clf_name in disp.line_.get_label()
pyplot.close("all")
clf_name = "another_name"
disp.plot(name=clf_name)
assert clf_name in disp.line_.get_label()
def test_precision_recall_curve_string_labels(pyplot):
# regression test #15738
cancer = load_breast_cancer()
X = cancer.data
y = cancer.target_names[cancer.target]
lr = make_pipeline(StandardScaler(), LogisticRegression())
lr.fit(X, y)
for klass in cancer.target_names:
assert klass in lr.classes_
disp = plot_precision_recall_curve(lr, X, y)
y_pred = lr.predict_proba(X)[:, 1]
avg_prec = average_precision_score(y, y_pred, pos_label=lr.classes_[1])
assert disp.average_precision == pytest.approx(avg_prec)
assert disp.estimator_name == lr.__class__.__name__
def test_errors(pyplot):
X, y_multiclass = make_classification(n_classes=3,
n_samples=50,
n_informative=3,
random_state=0)
y_binary = y_multiclass == 0
# Unfitted classifer
binary_clf = DecisionTreeClassifier()
with pytest.raises(NotFittedError):
plot_precision_recall_curve(binary_clf, X, y_binary)
binary_clf.fit(X, y_binary)
multi_clf = DecisionTreeClassifier().fit(X, y_multiclass)
# Fitted multiclass classifier with binary data
msg = "DecisionTreeClassifier should be a binary classifier"
with pytest.raises(ValueError, match=msg):
plot_precision_recall_curve(multi_clf, X, y_binary)
reg = DecisionTreeRegressor().fit(X, y_multiclass)
msg = "DecisionTreeRegressor should be a binary classifier"
with pytest.raises(ValueError, match=msg):
plot_precision_recall_curve(reg, X, y_binary)