ROCCurveFitDetector<T, TInput, TOutput>

Models & Types · AiDotNet.FitDetectors

A detector that evaluates model fit quality using ROC (Receiver Operating Characteristic) curve analysis.

Inheritance: FitDetectorBase<T, TInput, TOutput> → ROCCurveFitDetector<T, TInput, TOutput>

For Beginners

This class helps determine how well your classification model performs. It uses something called an “ROC curve” which is a way to visualize how good your model is at distinguishing between positive cases (like “yes, this email is spam”) and negative cases (like “no, this email is not spam”).

The key metric used is called “AUC” (Area Under the Curve), which gives a single number between 0 and 1:

  • AUC near 1.0: Your model is excellent at classification
  • AUC near 0.5: Your model is no better than random guessing
  • AUC near 0.0: Your model is consistently wrong (which means it could be fixed by inverting its predictions)

This detector will tell you if your model is good, moderate, poor, or very poor based on this AUC value.

Constructors

Constructor Summary
ROCCurveFitDetector(ROCCurveFitDetectorOptions options) Creates a new instance of the ROC curve fit detector with optional custom configuration.

Methods

Method Summary
FitDetectorResult<T> DetectFit(ModelEvaluationData<T, TInput, TOutput> evaluationData) Analyzes the model’s performance data and determines how well the model fits the data.