Table of Contents

Interface IFitDetector<T, TInput, TOutput>

Namespace
AiDotNet.Interfaces
Assembly
AiDotNet.dll

Defines an interface for detecting how well a machine learning model fits the data.

public interface IFitDetector<T, TInput, TOutput>

Type Parameters

T

The numeric type used for calculations (e.g., double, float).

TInput
TOutput

Remarks

For Beginners: This interface helps determine if your model is learning properly or has problems.

When training machine learning models, three common problems can occur:

  1. Underfitting: The model is too simple and doesn't capture important patterns in the data.

    • Like using only a house's age to predict its price, ignoring size, location, etc.
    • Signs: Poor performance on both training and new data

  2. Overfitting: The model memorizes the training data instead of learning general patterns.

    • Like memorizing specific houses and their prices instead of understanding what makes houses valuable
    • Signs: Excellent performance on training data but poor performance on new data

  3. Good fit: The model captures the important patterns without memorizing noise.

    • Like understanding that location, size, and condition affect house prices
    • Signs: Good performance on both training data and new data

This interface provides methods to analyze your model's performance and detect which of these situations you're dealing with, so you can make appropriate adjustments.

Methods

DetectFit(ModelEvaluationData<T, TInput, TOutput>)

Analyzes model evaluation data to detect whether the model is underfitting, overfitting, or has a good fit.

FitDetectorResult<T> DetectFit(ModelEvaluationData<T, TInput, TOutput> evaluationData)

Parameters

evaluationData ModelEvaluationData<T, TInput, TOutput>

Data containing model performance metrics on training and validation datasets.

Returns

FitDetectorResult<T>

A FitDetectorResult object containing the detected fit type and relevant metrics.

Remarks

For Beginners: This method examines how your model performs on different data sets to diagnose problems.

The input parameter:

  • evaluationData: Contains information about how well your model performs on:
    • Training data (the data your model learned from)
    • Validation data (new data your model hasn't seen before)

The method compares these performances to determine if your model:

  • Is underfitting (performs poorly on both training and validation data)
  • Is overfitting (performs very well on training data but poorly on validation data)
  • Has a good fit (performs well on both training and validation data)

The returned FitDetectorResult contains:

  • The detected fit type (underfitting, overfitting, or good fit)
  • Relevant metrics that support this conclusion
  • Possibly recommendations for how to improve the model

This information helps you decide what to do next:

  • If underfitting: Try a more complex model or add more features
  • If overfitting: Try regularization, get more training data, or simplify the model
  • If good fit: Your model is ready to use!
Edit this page