com.workfusion.vds.sdk.ml.sklearn.sgd

Class StochasticGradientDescentClassifier.Builder

java.lang.Object

com.workfusion.vds.sdk.ml.python.BasePythonClassifier.Builder<StochasticGradientDescentClassifier.Builder,StochasticGradientDescentClassifier>

com.workfusion.vds.sdk.ml.sklearn.sgd.StochasticGradientDescentClassifier.Builder

Enclosing class:

StochasticGradientDescentClassifier

public static class StochasticGradientDescentClassifier.Builder
extends BasePythonClassifier.Builder<StochasticGradientDescentClassifier.Builder,StochasticGradientDescentClassifier>

The type Builder.

Field Summary

Fields inherited from class com.workfusion.vds.sdk.ml.python.BasePythonClassifier.Builder

modelName, workingDir

Constructor Summary

Constructors

Constructor and Description
Builder() Instantiates a new Builder.

Method Summary

Modifier and Type	Method and Description
StochasticGradientDescentClassifier.Builder	alpha(double alpha) Sets the constant that multiplies the regularization term.
StochasticGradientDescentClassifier.Builder	average(boolean average) Sets whether computes the averaged SGD weights.
StochasticGradientDescentClassifier.Builder	average(int average) Sets the average.
protected StochasticGradientDescentClassifier	buildModel() Build classifier.
StochasticGradientDescentClassifier.Builder	classWeight(ClassWeightType classWeight) Sets class weight.
protected com.workfusion.automl.ml.bridge.python.classifier.PythonClassifierBridge	createDefaultClassifierBridge() Create default python classifier bridge.
StochasticGradientDescentClassifier.Builder	epsilon(double epsilon) Sets epsilon in the epsilon-insensitive loss functions.
StochasticGradientDescentClassifier.Builder	eta0(double eta0) Sets the initial learning rate.
StochasticGradientDescentClassifier.Builder	fitIntercept(boolean fitIntercept) Sets whether to calculate the intercept for this model.
StochasticGradientDescentClassifier.Builder	jobs(int jobs) Sets the number of jobs to run in parallel for both fit and predict.
StochasticGradientDescentClassifier.Builder	l1Ratio(double l1Ratio) Sets the Elastic Net mixing parameter [0, 1].
StochasticGradientDescentClassifier.Builder	learningRate(LearningRateType learningRate) Sets learning rate schedule.
StochasticGradientDescentClassifier.Builder	lossFunction(LossFunctionType lossFunction) Sets the loss function.
StochasticGradientDescentClassifier.Builder	maxIterations(int maxIterations) Sets the maximum number of iterations to be run.
StochasticGradientDescentClassifier.Builder	modelName(String name) Set a model name.
StochasticGradientDescentClassifier.Builder	penalty(PenaltyType penalty) Sets the norm used in the penalization.
StochasticGradientDescentClassifier.Builder	powerT(double powerT) Sets the exponent for inverse scaling learning rate.
StochasticGradientDescentClassifier.Builder	seed(Integer seed) Sets the seed used by the random number generator.
StochasticGradientDescentClassifier.Builder	shuffle(boolean shuffle) Sets whether or not the training data should be shuffled after each epoch.
StochasticGradientDescentClassifier.Builder	tolerance(Double tolerance) Sets the tolerance for stopping criteria.
StochasticGradientDescentClassifier.Builder	verbose(boolean verbose) Sets whether verbose mode is used.
StochasticGradientDescentClassifier.Builder	warmStart(boolean warmStart) Sets whether warm start is used.

Methods inherited from class com.workfusion.vds.sdk.ml.python.BasePythonClassifier.Builder

build, workingDir

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Builder

public Builder()

Instantiates a new Builder.

Method Detail

createDefaultClassifierBridge

protected com.workfusion.automl.ml.bridge.python.classifier.PythonClassifierBridge createDefaultClassifierBridge()

Description copied from class: BasePythonClassifier.Builder

Create default python classifier bridge.

Specified by:

createDefaultClassifierBridge in class BasePythonClassifier.Builder<StochasticGradientDescentClassifier.Builder,StochasticGradientDescentClassifier>

Returns:

the python classifier bridge

lossFunction

public StochasticGradientDescentClassifier.Builder lossFunction(LossFunctionType lossFunction)

Sets the loss function.

Parameters:

lossFunction - the criterion (LossFunctionType) (default: LossFunctionType.HINGE)

Returns:

the builder

penalty

public StochasticGradientDescentClassifier.Builder penalty(PenaltyType penalty)

Sets the norm used in the penalization.

Parameters:

penalty - the penalty (PenaltyType) (default: PenaltyType.L2)

Returns:

the builder

alpha

public StochasticGradientDescentClassifier.Builder alpha(double alpha)

Sets the constant that multiplies the regularization term. Also used to compute learningRate when set to LearningRateType.OPTIMAL.

Parameters:

alpha - the alpha (default: 1e-4)

Returns:

the builder

l1Ratio

public StochasticGradientDescentClassifier.Builder l1Ratio(double l1Ratio)

Sets the Elastic Net mixing parameter [0, 1]. 0 corresponds to L2 penalty, 1 to L1.

Parameters:

l1Ratio - the ratio parameter (default: 0.15)

Returns:

the builder

fitIntercept

public StochasticGradientDescentClassifier.Builder fitIntercept(boolean fitIntercept)

Sets whether to calculate the intercept for this model. If set to false, no intercept will be used in calculations (i.e. data is expected to be already centered).

Parameters:

fitIntercept - the fit intercept flag (default: true)

Returns:

the builder

maxIterations

public StochasticGradientDescentClassifier.Builder maxIterations(int maxIterations)

Sets the maximum number of iterations to be run.

Parameters:

maxIterations - the number of iterations (default: 5)

Returns:

the builder

tolerance

public StochasticGradientDescentClassifier.Builder tolerance(Double tolerance)

Sets the tolerance for stopping criteria. If it is not null, the iterations will stop when (loss > previous_loss - tolerance).

Parameters:

tolerance - the tolerance (default: null)

Returns:

the builder

shuffle

public StochasticGradientDescentClassifier.Builder shuffle(boolean shuffle)

Sets whether or not the training data should be shuffled after each epoch.

Parameters:

shuffle - the shuffle flag (default: true)

Returns:

the builder

verbose

public StochasticGradientDescentClassifier.Builder verbose(boolean verbose)

Sets whether verbose mode is used.

Parameters:

verbose - the verbose mode flag (default: false)

Returns:

the builder

epsilon

public StochasticGradientDescentClassifier.Builder epsilon(double epsilon)

Sets epsilon in the epsilon-insensitive loss functions.

Parameters:

epsilon - the epsilon (default: 0.1)

Returns:

the builder

jobs

public StochasticGradientDescentClassifier.Builder jobs(int jobs)

Sets the number of jobs to run in parallel for both fit and predict. If -1, then the number of jobs is set to the number of cores.

Parameters:

jobs - the number of jobs (default: 1)

Returns:

the builder

seed

public StochasticGradientDescentClassifier.Builder seed(Integer seed)

Sets the seed used by the random number generator. If null then default random number generator is used (np.random).

Parameters:

seed - the seed (default: null)

Returns:

the builder

learningRate

public StochasticGradientDescentClassifier.Builder learningRate(LearningRateType learningRate)

Sets learning rate schedule.

Parameters:

learningRate - the learning rate (LearningRateType) (default: LearningRateType.OPTIMA)

Returns:

the builder

eta0

public StochasticGradientDescentClassifier.Builder eta0(double eta0)

Sets the initial learning rate.

Parameters:

eta0 - the initial learning rate (default: 0.0)

Returns:

the builder

powerT

public StochasticGradientDescentClassifier.Builder powerT(double powerT)

Sets the exponent for inverse scaling learning rate.

Parameters:

powerT - the exponent (default: 0.5)

Returns:

the builder

classWeight

public StochasticGradientDescentClassifier.Builder classWeight(ClassWeightType classWeight)

Sets class weight. If null then all classes are supposed to have weight one.

Parameters:

classWeight - the class weight (ClassWeightType) (default: null)

Returns:

the builder

warmStart

public StochasticGradientDescentClassifier.Builder warmStart(boolean warmStart)

Sets whether warm start is used.

Parameters:

warmStart - the warm start (default: false)

Returns:

the builder

average

public StochasticGradientDescentClassifier.Builder average(int average)

Sets the average. Averaging will begin once the total number of samples seen reaches average.

Parameters:

average - the average weights

Returns:

the builder

average

public StochasticGradientDescentClassifier.Builder average(boolean average)

Sets whether computes the averaged SGD weights.

Parameters:

average - the average flag (default: false)

Returns:

the builder

modelName

public StochasticGradientDescentClassifier.Builder modelName(String name)

Description copied from class: BasePythonClassifier.Builder

Set a model name.

Overrides:

modelName in class BasePythonClassifier.Builder<StochasticGradientDescentClassifier.Builder,StochasticGradientDescentClassifier>

Parameters:

name - the model name (String)

Returns:

the builder

buildModel

protected StochasticGradientDescentClassifier buildModel()

Description copied from class: BasePythonClassifier.Builder

Build classifier.

Specified by:

buildModel in class BasePythonClassifier.Builder<StochasticGradientDescentClassifier.Builder,StochasticGradientDescentClassifier>

Returns:

the classifier instance