SVM.LinearSVR

Linear support vector regression trained with gradient descent on the epsilon-insensitive loss.

interface LinearSVRProps {
    epsilon?: number;
    C?: number;
    maxIter?: number;
    learningRate?: number;
}
constructor(props: LinearSVRProps = {})

Parameters

• epsilon (number, default 0): width of the insensitive tube around the regression line
• C (number, default 1): regularization strength
• maxIter (number, default 100): maximum training iterations
• learningRate (number, default 0.01): optimizer step size

Example

const reg = new LinearSVR();
reg.fit(X, y);
const preds = reg.predict(T);