def test_make_imbalanced_iris(as_frame, sampling_strategy, expected_counts):
pytest.importorskip("pandas")
X, y = fetch_openml("iris", version=1, return_X_y=True, as_frame=as_frame)
X_res, y_res = make_imbalance(X, y, sampling_strategy=sampling_strategy)
if as_frame:
assert hasattr(X_res, "loc")
assert Counter(y_res) == expected_counts
def check_classifiers_with_encoded_labels(name, classifier_orig):
# Non-regression test for #709
# https://github.com/scikit-learn-contrib/imbalanced-learn/issues/709
pytest.importorskip("pandas")
classifier = clone(classifier_orig)
df, y = fetch_openml("iris", version=1, as_frame=True, return_X_y=True)
df, y = make_imbalance(
df,
y,
sampling_strategy={
"Iris-setosa": 30,
"Iris-versicolor": 20,
"Iris-virginica": 50,
},
)
classifier.set_params(sampling_strategy={
"Iris-setosa": 20,
"Iris-virginica": 20
})
classifier.fit(df, y)
assert set(classifier.classes_) == set(y.cat.categories.tolist())
y_pred = classifier.predict(df)
assert set(y_pred) == set(y.cat.categories.tolist())
def test_make_imbalance_dict(iris, sampling_strategy, expected_counts):
X, y = iris
_, y_ = make_imbalance(X, y, sampling_strategy=sampling_strategy)
assert Counter(y_) == expected_counts
def test_make_imbalance_error_single_class(iris):
X, y = iris
y = np.zeros_like(y)
with pytest.raises(ValueError, match="needs to have more than 1 class."):
make_imbalance(X, y, sampling_strategy={0: 10})
def test_make_imbalance_error(iris, sampling_strategy, err_msg):
# we are reusing part of utils.check_sampling_strategy, however this is not
# cover in the common tests so we will repeat it here
X, y = iris
with pytest.raises(ValueError, match=err_msg):
make_imbalance(X, y, sampling_strategy=sampling_strategy)
import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
RANDOM_STATE = 42
# %% [markdown]
# Preparation
# -----------
# **Make 3 imbalanced toy classification tasks.**
distribution = {0: 100, 1: 50}
# dataset 1
X, y = make_moons(200, noise=0.2, random_state=RANDOM_STATE)
dataset1 = make_imbalance(X,
y,
sampling_strategy=distribution,
random_state=RANDOM_STATE)
# dataset 2
X, y = make_circles(200, noise=0.2, factor=0.5, random_state=RANDOM_STATE)
dataset2 = make_imbalance(X,
y,
sampling_strategy=distribution,
random_state=RANDOM_STATE)
# dataset 3
X, y = make_classification(200,
n_features=2,
n_redundant=0,
n_informative=2,
random_state=1,
n_clusters_per_class=1)
X += 2 * np.random.RandomState(RANDOM_STATE).uniform(size=X.shape)
imbalance_distr = {
0: 178,
1: 120,
2: 80,
3: 60,
4: 50,
5: 44,
6: 40,
7: 40,
8: 40,
9: 40
}
X, y = make_imbalance(X,
y,
sampling_strategy=imbalance_distr,
random_state=RANDOM_STATE)
fig = plot_2Dprojection_and_cardinality(X, y, figsize=(8, 4))
# %% [markdown]
# Classification
# --------------
# We split the data into train and test subsets and fit a ``SelfPacedEnsembleClassifier`` (with support vector machine as base classifier) on the train samples.
# The fitted classifier can subsequently be used to predict the value of the digit for the samples in the test subset.
# Split data into 50% train and 50% test subsets
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.5,
shuffle=True,
RANDOM_STATE = 42
# %% [markdown]
# Preparation
# -----------
# **Make 3 imbalanced iris classification tasks.**
iris = sklearn.datasets.load_iris()
X = iris.data[:, 0:2] # we only take the first two features for visualization
y = iris.target
X, y = make_imbalance(X,
y,
sampling_strategy={
0: 50,
1: 30,
2: 10
},
random_state=RANDOM_STATE)
print('Class distribution of imbalanced iris dataset: \n%s' %
sort_dict_by_key(Counter(y)))
# %% [markdown]
# **Create SPE (ensemble size = 5) with different base classifiers.**
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.gaussian_process import GaussianProcessClassifier
from sklearn.gaussian_process.kernels import RBF
classifiers = {
data=X,
x="feature 1",
y="feature 2",
hue=y,
style=y,
ax=axs[0, 0],
)
axs[0, 0].set_title("Original set")
multipliers = [0.9, 0.75, 0.5, 0.25, 0.1]
for ax, multiplier in zip(axs.ravel()[1:], multipliers):
X_resampled, y_resampled = make_imbalance(
X,
y,
sampling_strategy=ratio_func,
**{
"multiplier": multiplier,
"minority_class": 1
},
)
sns.scatterplot(
data=X_resampled,
x="feature 1",
y="feature 2",
hue=y_resampled,
style=y_resampled,
ax=ax,
)
ax.set_title(f"Sampling ratio = {multiplier}")