MatthewsCorrelationCoefficientMetric<T>

Models & Types · AiDotNet.Evaluation.Metrics.Classification

Computes Matthews Correlation Coefficient (MCC): a balanced measure for binary and multi-class classification.

Implements: IClassificationMetric<T>

For Beginners

MCC is one of the best single-number metrics for classification, especially with imbalanced classes. It uses all four quadrants of the confusion matrix and produces a high score only when the model does well on both positives and negatives.

How It Works

For binary classification: MCC = (TPTN - FPFN) / sqrt((TP+FP)(TP+FN)(TN+FP)(TN+FN))

Interpretation:

  • MCC = +1: Perfect prediction
  • MCC = 0: No better than random
  • MCC = -1: Complete disagreement (predictions are inverted)

Advantages over other metrics:

  • Works well with imbalanced classes (unlike accuracy)
  • Considers all parts of confusion matrix (unlike F1)
  • Is symmetric: swapping positives/negatives gives same |MCC|

Research reference: Matthews (1975), Chicco & Jurman (2020) recommend MCC over F1.

Constructors

Constructor Summary
MatthewsCorrelationCoefficientMetric() Initializes a new MCC metric with default positive label (1).
MatthewsCorrelationCoefficientMetric(T positiveLabel) Initializes a new MCC metric with an explicit positive label.

Properties

Property Summary
MetricDirection Direction { get; }
T MaxValue { get; }
T MinValue { get; }
bool RequiresProbabilities { get; }
bool SupportsMultiClass { get; }
string Category { get; }
string Description { get; }
string Name { get; }

Methods

Method Summary
MetricWithCI<T> ComputeWithCI(ReadOnlySpan<T> predictions, ReadOnlySpan<T> actuals, ConfidenceIntervalMethod ciMethod, double confidenceLevel, int bootstrapSamples, int? randomSeed)
T Compute(ReadOnlySpan<T> predictions, ReadOnlySpan<T> actuals)