SVMRegressionTrainEpsilonSVR Method

public void TrainEpsilonSVR(
	Matrix observations,
	Vector values,
	bool autoScale = true,
	SVMKernelBase kernel = null,
	double C = 1,
	double epsilon = 0.1,
	double cacheSize = 100,
	double eps = 0.001,
	int maximumIterations = 0,
	TimeSpan modelTrainTimeout = null,
	bool shrinkingHeuristics = true
)

Parameters

observations

Type: FinMath.LinearAlgebraMatrix
Observation matrix (N*M size). Columns corresponds to factors, rows to observation.

values

Type: FinMath.LinearAlgebraVector
The vector (N size) predicted values.

autoScale (Optional)

Type: SystemBoolean
Auto scale observations. Normalize the observations matrix such that columns of observations are centered to have mean 0 and scaled to have standard deviation 1.

kernel (Optional)

Type: FinMath.MachineLearning.SVMSVMKernelBase
The kernel function for SVM model (default RadialBasisFunction).

C (Optional)

Type: SystemDouble
C is the penalty parameter of the error term (default 1).

epsilon (Optional)

Type: SystemDouble
The epsilon in loss function (default 0.1)

cacheSize (Optional)

Type: SystemDouble
The cache memory size in MB (default 100).

eps (Optional)

Type: SystemDouble
The tolerance of termination criterion (default 0.001).

maximumIterations (Optional)

Type: SystemInt32
The maximum iterations number (0 means auto).

modelTrainTimeout (Optional)

Type: SystemTimeSpan
One model train timeout (0 means no restriction).

shrinkingHeuristics (Optional)

Type: SystemBoolean
Whether to use the shrinking heuristics flag (default true).

Reference