predict.cv.kmr: Make prediction with a "'cv.kmr'" object

In jpvert/kmr: Kernel multitask regression

Description

Similar to other predict methods, this function predicts fitted values from a fitted "cv.kmr" object.

Usage

## S3 method for class 'cv.kmr'
predict(object, newx, lambda = "lambda.opt", ...)

Arguments

object	Fitted "cv.kmr" model object.
newx	Matrix of new values for x or kernel matrix for new values crossing old values for x, at which predictions are to be made.
lambda	Value(s) of the regularization parameter, a single scalar or a sequence of values, or "lambda.opt" which allows different lambdas for each task tuned by cross-validation. Default is "lambda.opt".
...	Not used. Other arguments to predict.

Value

A matrix of predicted values for the new samples (in rows) and all tasks (in columns), corresponding to the regularization parameter lambda. If lambda is a list of specified values, the function returns a list of matrices, corresponding to the predictions for the different values of lambda in the list. If lambda is "lambda.opt", a single matrix returned too.

Note

Note that "lambda.opt" will allow different lambdas for each task tuned by cross-validation; otherwise, specified value(s) of lambda will be fixed for all tasks.

See Also

cv.kmr

jpvert/kmr documentation built on May 20, 2019, 7:56 a.m.