Class LinearClassifierOptions<T>

Namespace

AiDotNet.Models.Options

Assembly

AiDotNet.dll

Configuration options for linear classifiers.

public class LinearClassifierOptions<T> : ClassifierOptions<T>

Type Parameters

T

The data type used for calculations.

Inheritance

object

ModelOptions

ClassifierOptions<T>

LinearClassifierOptions<T>

Derived

PassiveAggressiveOptions<T>

Inherited Members

ClassifierOptions<T>.TaskType

ClassifierOptions<T>.DecisionThreshold

ClassifierOptions<T>.UseClassWeights

ClassifierOptions<T>.ClassWeights

ModelOptions.Seed

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

Linear classifiers learn a linear decision boundary to separate classes. They are simple, interpretable, and often very effective.

For Beginners: Linear classifiers draw a straight line (or hyperplane in higher dimensions) to separate different classes of data.

Key concepts:

• They learn weights for each feature
• Prediction is: sign(weight * features + bias)
• Training adjusts weights to minimize errors

Linear classifiers are great when:

• You have many features (high-dimensional data)
• Data is approximately linearly separable
• You need fast training and prediction
• You want interpretable models (feature importance = weight magnitude)

Properties

Alpha

Gets or sets the regularization strength (alpha).

public double Alpha { get; set; }

Property Value

double

A positive regularization parameter. Default is 0.0001.

Remarks

Higher values mean stronger regularization, which prevents overfitting but may increase bias. Set to 0 for no regularization.

FitIntercept

Gets or sets whether to fit an intercept term (bias).

public bool FitIntercept { get; set; }

Property Value

bool

True (default) to include an intercept; false to pass through origin.

Remarks

For Beginners: The intercept (also called bias) allows the decision boundary to not pass through the origin.

Almost always set to true unless you specifically know your data should be classified through the origin.

LearningRate

Gets or sets the learning rate for gradient-based optimization.

public double LearningRate { get; set; }

Property Value

double

A positive learning rate. Default is 0.01.

Remarks

Controls how much to adjust weights on each update. - Too high: Training may oscillate or diverge - Too low: Training will be slow to converge

Loss

Gets or sets the loss function type.

public LinearLoss Loss { get; set; }

Property Value

LinearLoss

The loss function. Default is Hinge (for SVM-style classifier).

MaxIterations

Gets or sets the maximum number of training iterations.

public int MaxIterations { get; set; }

Property Value

int

The maximum iterations. Default is 1000.

Penalty

Gets or sets the penalty type for regularization.

public LinearPenalty Penalty { get; set; }

Property Value

LinearPenalty

The penalty type. Default is L2.

RandomState

Gets or sets the random state for reproducibility.

public int? RandomState { get; set; }

Property Value

int?

The random seed, or null for non-deterministic. Default is null.

Shuffle

Gets or sets whether to shuffle training data at each epoch.

public bool Shuffle { get; set; }

Property Value

bool

True (default) to shuffle; false to use data in original order.

Tolerance

Gets or sets the convergence tolerance.

public double Tolerance { get; set; }

Property Value

double

The tolerance for convergence check. Default is 1e-4.

Remarks

Training stops early if the loss improvement is below this threshold.