UncertaintyPredictionResult<T, TOutput>

Models & Types · AiDotNet.Models.Results

Represents a prediction result augmented with uncertainty information.

For Beginners

This lets you ask the model both:

  • “What is the prediction?”
  • “How uncertain is that prediction?”

How It Works

This type is returned by the facade method AiModelResult.PredictWithUncertainty(...).

Constructors

Constructor Summary
UncertaintyPredictionResult(UncertaintyQuantificationMethod methodUsed, TOutput prediction, TOutput variance, IReadOnlyDictionary<string, Tensor<T>> metrics, RegressionConformalInterval<TOutput> regressionInterval, ClassificationConformalPredictionSet classificationSet) Initializes a new instance of the UncertaintyPredictionResult class.

Properties

Property Summary
ClassificationConformalPredictionSet ClassificationSet { get; } Gets an optional conformal classification prediction set, when configured and supported.
IReadOnlyDictionary<string, Tensor<T>> Metrics { get; } Gets additional uncertainty diagnostics.
RegressionConformalInterval<TOutput> RegressionInterval { get; } Gets an optional conformal regression interval, when configured and supported.
TOutput Prediction { get; } Gets the point prediction (mean / expected prediction).
TOutput Variance { get; } Gets the per-output predictive variance (when available).
UncertaintyQuantificationMethod MethodUsed { get; } Gets the uncertainty method that was used to produce this result.