Class HybridFitDetectorOptions

Namespace

AiDotNet.Models.Options

Assembly

AiDotNet.dll

Configuration options for the Hybrid Fit Detector, which combines multiple model evaluation techniques to provide a comprehensive assessment of model quality.

public class HybridFitDetectorOptions

Inheritance

object

HybridFitDetectorOptions

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

The Hybrid Fit Detector uses a combination of approaches to evaluate model fit, including comparing training and validation performance, analyzing residuals, and examining model complexity. This comprehensive approach provides a more robust assessment than any single method alone, helping to identify overfitting, underfitting, and good fit conditions with greater confidence.

For Beginners: This detector is like having a team of experts evaluate your machine learning model from different angles. Instead of relying on just one way to check if your model is learning properly, it uses several different methods and combines their insights.

Think of it like a comprehensive health checkup that includes multiple tests (blood work, physical exam, imaging, etc.) rather than just checking your temperature. By looking at your model from multiple perspectives, the Hybrid Fit Detector can give you a more complete picture of how well your model is learning and where it might be having problems.

The detector helps identify three common scenarios:

• Overfitting: Your model has "memorized" the training data but doesn't generalize well to new data
• Underfitting: Your model is too simple and isn't capturing important patterns in the data
• Good Fit: Your model has found the right balance, learning meaningful patterns that generalize well

This hybrid approach is especially useful when individual detection methods might give conflicting signals or when you want extra confidence in your model quality assessment.

Properties

GoodFitThreshold

Gets or sets the threshold for confirming good fit based on a composite score from multiple evaluation methods.

public double GoodFitThreshold { get; set; }

Property Value

double

The good fit threshold, defaulting to 0.8 (80%).

Remarks

This threshold determines when a model is considered to have a good fit based on a composite score that combines multiple evaluation metrics. The exact calculation depends on the implementation, but typically includes factors like balanced performance between training and validation data, appropriate model complexity for the problem, and the absence of concerning patterns in the residuals.

For Beginners: This setting helps identify when your model has achieved the right balance - learning meaningful patterns without memorizing the training data. With the default value of 0.8, if the composite good fit score exceeds 80%, your model is considered to have a good fit.

The composite score looks at several indicators of good fit, such as:

• Strong performance on both training and validation data, with the gap between them not being too large
• Appropriate model complexity for the problem and amount of data
• Residuals (errors) that look random rather than showing patterns
• Consistent performance across different subsets of the data

For example, a model with good validation performance might not be flagged as having a good fit if there are other concerning signs like highly non-random errors. Conversely, a model with slightly lower performance but very clean residuals and appropriate complexity might be recognized as having a good fit.

A good fit means your model has struck the right balance between underfitting and overfitting - it's complex enough to learn from the data but not so complex that it just memorizes it. This is the "Goldilocks zone" we aim for in machine learning.

OverfitThreshold

Gets or sets the threshold for detecting overfitting based on a composite score from multiple evaluation methods.

public double OverfitThreshold { get; set; }

Property Value

double

The overfit threshold, defaulting to 0.2 (20%).

Remarks

This threshold determines when a model is considered to be overfitting based on a composite score that combines multiple evaluation metrics. The exact calculation depends on the implementation, but typically includes factors like the gap between training and validation performance, the complexity of the model relative to the amount of training data, and patterns in the residuals.

For Beginners: This setting helps identify when your model is "memorizing" the training data instead of learning general patterns. With the default value of 0.2, if the composite overfitting score exceeds 20%, your model is flagged as overfitting.

The composite score looks at several warning signs of overfitting, such as:

• How much better your model performs on training data than on validation data
• Whether your model is unnecessarily complex for the amount of data you have
• If your model's errors show suspicious patterns

For example, a model might have only a small gap between training and validation performance (which alone wouldn't trigger an overfitting warning), but if it's also very complex and shows certain patterns in its errors, the combined evidence might push the composite score above the threshold, indicating overfitting.

When overfitting is detected, you might want to:

• Use more regularization to penalize complexity
• Reduce model complexity (fewer features, simpler model)
• Gather more training data
• Use techniques like early stopping or dropout

UnderfitThreshold

Gets or sets the threshold for detecting underfitting based on a composite score from multiple evaluation methods.

public double UnderfitThreshold { get; set; }

Property Value

double

The underfit threshold, defaulting to 0.5 (50%).

Remarks

This threshold determines when a model is considered to be underfitting based on a composite score that combines multiple evaluation metrics. The exact calculation depends on the implementation, but typically includes factors like the absolute performance on both training and validation data, the simplicity of the model relative to the complexity of the problem, and patterns in the residuals.

For Beginners: This setting helps identify when your model is "not learning enough" from the training data. With the default value of 0.5, if the composite underfitting score exceeds 50%, your model is flagged as underfitting.

The composite score looks at several warning signs of underfitting, such as:

• Poor performance on both training and validation data
• Whether your model is too simple for the complexity of the problem
• If your model's errors show patterns that suggest missed relationships

For example, a model might have mediocre performance that alone wouldn't definitively indicate underfitting, but if it's also very simple and shows strong patterns in its errors that suggest missed relationships, the combined evidence might push the composite score above the threshold, indicating underfitting.

When underfitting is detected, you might want to:

• Increase model complexity (more features, more complex model)
• Train for more iterations or epochs
• Reduce regularization strength
• Engineer better features
• Try a different type of model altogether