Class AutoMLEnsembleOptions

Namespace

AiDotNet.Configuration

Assembly

AiDotNet.dll

Configuration options for AutoML ensembling.

public sealed class AutoMLEnsembleOptions

Inheritance

object

AutoMLEnsembleOptions

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

Ensembling retrains a small set of top-performing trials on the full training data and combines their predictions into a single "ensemble" model. This can improve accuracy and reduce variance.

For Beginners: An ensemble is like asking multiple experts and averaging their answers. It often performs better than relying on a single model.

Properties

Enabled

Gets or sets a value indicating whether AutoML should attempt to build an ensemble after the search.

public bool Enabled { get; set; }

Property Value

bool

FinalSelectionPolicy

Gets or sets the policy that determines whether the ensemble replaces the best single model.

public AutoMLFinalModelSelectionPolicy FinalSelectionPolicy { get; set; }

Property Value

AutoMLFinalModelSelectionPolicy

MaxModelCount

Gets or sets the maximum number of top trials to include in the ensemble.

public int MaxModelCount { get; set; }

Property Value

int

Remarks

Values less than 2 disable ensembling (an ensemble requires at least two models).