HeteroscedasticityFitDetector<T, TInput, TOutput>
Models & Types · AiDotNet.FitDetectors
A detector that evaluates whether a model’s errors have consistent variance across all predictions.
Inheritance: FitDetectorBase<T, TInput, TOutput> → HeteroscedasticityFitDetector<T, TInput, TOutput>
For Beginners
Heteroscedasticity is a statistical term that means “uneven spread” of errors. In a good model, the errors (differences between predictions and actual values) should be roughly the same size regardless of what you’re predicting. If errors get much larger or smaller for certain predictions (like having more accurate predictions for small values but less accurate for large values), that’s called heteroscedasticity, and it can make your model less reliable.
This detector helps you identify if your model has this problem and suggests ways to fix it.
Constructors
| Constructor | Summary |
|---|---|
HeteroscedasticityFitDetector(HeteroscedasticityFitDetectorOptions options) |
Initializes a new instance of the HeteroscedasticityFitDetector<T> class. |
Methods
| Method | Summary |
|---|---|
FitDetectorResult<T> DetectFit(ModelEvaluationData<T, TInput, TOutput> evaluationData) |
Analyzes model performance data to determine if the model has consistent error variance. |