Temperature Scaling Calibration
Scikit-learn 1.8 adds the "temperature" method to CalibratedClassifierCV. Temperature scaling is a simple yet effective technique for calibrating classifier probabilities: a single scalar parameter is learned to adjust the softmax distribution.
Practical Example
python
from sklearn.datasets import make_classification
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.calibration import CalibratedClassifierCV
from sklearn.model_selection import train_test_split
X, y = make_classification(n_samples=5000, random_state=42)
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.3, random_state=42
)
# Base classifier
clf = GradientBoostingClassifier(n_estimators=100)
# Temperature scaling calibration
calibrated = CalibratedClassifierCV(
clf, method='temperature', cv=5
)
calibrated.fit(X_train, y_train)
# Better calibrated probabilities
proba = calibrated.predict_proba(X_test)
print(f'Shape: {proba.shape}') # (1500, 2)