Class RegressorConfig
Configuration for regressor creation.
Implements
Inherited Members
Namespace: MOSAIC.Components.MachineLearning.Factory
Assembly: MOSAIC.dll
Syntax
public sealed record RegressorConfig : IEquatable<RegressorConfig>
Properties
Delta
RLS initialization regularization (delta). P0 = (1/delta) I. Only used for RecursiveLeastSquares. Default: 1.0.
Declaration
public double Delta { get; init; }
Property Value
| Type | Description |
|---|---|
| double |
FeatureDim
Number of random Fourier features (D). Only used for RFF. Default: 300. Higher = better kernel approximation but slower.
Declaration
public int FeatureDim { get; init; }
Property Value
| Type | Description |
|---|---|
| int |
ForgettingFactor
RLS forgetting factor ff in (0,1]. 1.0 = no forgetting. Only used for RecursiveLeastSquares. Default: 1.0. Typical for drift adaptation: 0.99..0.999
Declaration
public double ForgettingFactor { get; init; }
Property Value
| Type | Description |
|---|---|
| double |
Lambda
Regularization parameter (λ). Default: 1.0. Higher values = stronger regularization = simpler model.
Declaration
public double Lambda { get; init; }
Property Value
| Type | Description |
|---|---|
| double |
Seed
Random seed for RFF. Null = random. Set for reproducibility.
Declaration
public int? Seed { get; init; }
Property Value
| Type | Description |
|---|---|
| int? |
Sigma
RBF kernel bandwidth (σ). Only used for RFF. Default: 1.0. Controls how "local" the kernel is.
Declaration
public double Sigma { get; init; }
Property Value
| Type | Description |
|---|---|
| double |