CalibratedClassifier<T>

Models & Types · AiDotNet.Classification.Calibration

Wrapper that adds probability calibration to any probabilistic classifier.

Inheritance: ClassifierBase<T> → ProbabilisticClassifierBase<T> → CalibratedClassifier<T>

Implements: IGeneratedParameterRegistrar<T>

For Beginners

Many classifiers give poor probability estimates:

  • Random Forest tends to push probabilities towards 0.5
  • SVM’s margin-based scores aren’t true probabilities
  • Neural networks without proper training can be overconfident

This wrapper fixes that by learning to transform raw scores into well-calibrated probabilities that match actual event frequencies.

Example: If the model says “80% probability”, approximately 80% of such predictions should actually be positive.

Usage:

How It Works

CalibratedClassifier wraps any probabilistic classifier and applies post-hoc probability calibration to improve the reliability of predicted probabilities.

Constructors

Constructor Summary
CalibratedClassifier() Initializes a new instance with default settings using Gaussian Naive Bayes as the base classifier.
CalibratedClassifier(IProbabilisticClassifier<T> baseClassifier, CalibratedClassifierOptions<T> options, IRegularization<T, Matrix<T>, Vector<T>> regularization) Initializes a new CalibratedClassifier.

Properties

Property Summary
IProbabilisticClassifier<T> BaseClassifier { get; } Gets the base classifier.
ProbabilityCalibrationMethod CalibrationMethod { get; } Gets the calibration method.
bool IsTrained { get; } Gets whether the model is trained.

Methods

Method Summary
IFullModel<T, Matrix<T>, Vector<T>> WithParameters(Vector<T> parameters)
Matrix<T> PredictProbabilities(Matrix<T> input) Gets calibrated probability predictions.
ModelMetadata<T> GetModelMetadata()
void Train(Matrix<T> x, Vector<T> y) Trains the base classifier and fits calibration.