linear_model.SGDRegressor

Usage

import { SGDRegressor } from 'machinelearn/linear_model';
const reg = new SGDRegressor();
const X = [[0., 0.], [1., 1.]];
const y = [0, 1];
reg.fit(X, y);
reg.predict([[2., 2.]]); // result: [ 1.281828588248001 ]

Constructors

constructor

⊕ SGDRegressor(__namedParameters: `object`)

Defined in linear_model/stochastic_gradient.ts:40

Parameters:

Param	Type	Default
options.clone	boolean	true
options.epochs	number	10000
options.learning_rate	number	0.0001
options.loss	string	TypeLoss.L2
options.random_state	number	null

Returns: SGDRegressor

Properties

▸ epochs

Defined in linear_model/stochastic_gradient.ts:27

▸ learningRate

Defined in linear_model/stochastic_gradient.ts:26

▸ loss

Defined in linear_model/stochastic_gradient.ts:28

▸ regFactor

Defined in linear_model/stochastic_gradient.ts:29

Methods

λ fit

Train the base SGD

Defined in linear_model/stochastic_gradient.ts:102

Parameters:

Param	Type	Default	Description
X	number[][]	null	Matrix of data
y	number[]	null	Matrix of targets

Returns:

void

λ fromJSON

Restore the model from a checkpoint

Defined in linear_model/stochastic_gradient.ts:151

Parameters:

Param	Type	Default
options.epochs	number	10000
options.learning_rate	number	0.0001
options.random_state	number	null
options.weights	number[]	[]

Returns:

void

λ predict

Predicted values

Defined in linear_model/stochastic_gradient.ts:297

Parameters:

Param	Type	Default	Description
X	number[][]	null	Matrix of data

Returns:

number[]

λ toJSON

Save the model's checkpoint

Defined in linear_model/stochastic_gradient.ts:118

Returns:

Param	Type	Description
epochs	number	model training epochs
learning_rate	number	model learning rate
random_state	number	Number used to set a static random state
weights	number[]	Model training weights