Multiclass classification with under-sampling

Some balancing methods allow for balancing dataset with multiples classes. We provide an example to illustrate the use of those methods which do not differ from the binary case.

Training target statistics: Counter({np.int64(1): 38, np.int64(2): 38, np.int64(0): 17})
Testing target statistics: Counter({np.int64(1): 12, np.int64(2): 12, np.int64(0): 8})

		LogisticRegression
Metric	Label / Average
Accuracy		0.812500
Precision	0	1.000000
	1	0.875000
	2	0.687500
Recall	0	1.000000
	1	0.583333
	2	0.916667
ROC AUC	0	1.000000
	1	0.958333
	2	0.966667
Log loss		0.374343
Fit time (s)		NaN
Predict time (s)		0.000601

# Authors: Guillaume Lemaitre <g.lemaitre58@gmail.com>
# License: MIT

from collections import Counter

import skore
from sklearn.datasets import load_iris
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler

from imblearn.datasets import make_imbalance
from imblearn.pipeline import make_pipeline
from imblearn.under_sampling import NearMiss

print(__doc__)

RANDOM_STATE = 42

# Create a folder to fetch the dataset
iris = load_iris()
X, y = make_imbalance(
    iris.data,
    iris.target,
    sampling_strategy={0: 25, 1: 50, 2: 50},
    random_state=RANDOM_STATE,
)

X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=RANDOM_STATE)

print(f"Training target statistics: {Counter(y_train)}")
print(f"Testing target statistics: {Counter(y_test)}")

# Create a pipeline
pipeline = make_pipeline(NearMiss(version=2), StandardScaler(), LogisticRegression())
pipeline.fit(X_train, y_train)

# Classify and report the results
report = skore.evaluate(pipeline, X_test, y_test, splitter="prefit")
report.metrics.summarize().frame()

Estimated memory usage: 313 MB