FitnessCalculatorBase<T, TInput, TOutput>

Base Classes · AiDotNet.FitnessCalculators

Base class for all fitness calculators that evaluate how well a model performs.

Implements: IPreferredDataSetFitnessCalculator<T, TInput, TOutput>

Known derived types: AdjustedRSquaredFitnessCalculator<T, TInput, TOutput>, BinaryCrossEntropyLossFitnessCalculator<T, TInput, TOutput>, CategoricalCrossEntropyLossFitnessCalculator<T, TInput, TOutput>, ContrastiveLossFitnessCalculator<T, TInput, TOutput>, CosineSimilarityLossFitnessCalculator<T, TInput, TOutput>, CrossEntropyLossFitnessCalculator<T, TInput, TOutput>, DiceLossFitnessCalculator<T, TInput, TOutput>, ElasticNetLossFitnessCalculator<T, TInput, TOutput>, ExponentialLossFitnessCalculator<T, TInput, TOutput>, FocalLossFitnessCalculator<T, TInput, TOutput>, HingeLossFitnessCalculator<T, TInput, TOutput>, HuberLossFitnessCalculator<T, TInput, TOutput>, JaccardLossFitnessCalculator<T, TInput, TOutput>, KullbackLeiblerDivergenceFitnessCalculator<T, TInput, TOutput>, LogCoshLossFitnessCalculator<T, TInput, TOutput>, MeanAbsoluteErrorFitnessCalculator<T, TInput, TOutput>, MeanSquaredErrorFitnessCalculator<T, TInput, TOutput>, ModifiedHuberLossFitnessCalculator<T, TInput, TOutput>, OrdinalRegressionLossFitnessCalculator<T, TInput, TOutput>, PoissonLossFitnessCalculator<T, TInput, TOutput>, QuantileLossFitnessCalculator<T, TInput, TOutput>, RSquaredFitnessCalculator<T, TInput, TOutput>, RootMeanSquaredErrorFitnessCalculator<T, TInput, TOutput>, SquaredHingeLossFitnessCalculator<T, TInput, TOutput>, TripletLossFitnessCalculator<T, TInput, TOutput>, WeightedCrossEntropyLossFitnessCalculator<T, TInput, TOutput>

For Beginners

This is a foundation class that all fitness calculators build upon.

Think of a fitness calculator like a judge in a competition:

  • It looks at how your AI model performed (its predictions vs. the actual answers)
  • It gives a score based on specific criteria (like accuracy, error rate, etc.)
  • It helps you determine if one model is better than another

Different fitness calculators judge models in different ways, just like different sports have different scoring systems. Some calculators consider higher scores better (like accuracy), while others consider lower scores better (like error rates).

This base class provides the common functionality that all these different “judges” share.

Properties

Property Summary
DataSetType PreferredDataSetType { get; }
bool IsHigherScoreBetter { get; } Gets a value indicating whether higher fitness scores represent better performance.

Methods

Method Summary
T CalculateFitnessScore(DataSetStats<T, TInput, TOutput> dataSet) Calculates the fitness score using a specific dataset.
T CalculateFitnessScore(ModelEvaluationData<T, TInput, TOutput> evaluationData) Calculates the fitness score using a specific dataset.
bool IsBetterFitness(T newScore, T currentBestScore) Determines whether a new fitness score is better than the current best score.