predict.cv.rgam: Make predictions from a "cv.rgam" object

Description

This function returns the predictions for a new data matrix from a cross-validated rgam model by using the stored "glmfit" object and the optimal value chosen for lambda.

Usage

# S3 method for cv.rgam
predict(object, xnew, s = c("lambda.1se",
  "lambda.min"), ...)

Arguments

object

Fitted "cv.rgam" object.

xnew

Matrix of new values for x at which predictions are to be made.

Value of the penalty parameter lambda at which predictions are required. Default is the value s="lambda.1se" stored in the CV fit. Alternatively, s="lambda.min" can be used. If s is numeric, it is taken as the value(s) of lambda to be used.

...

Other arguments to be passed to predict.rgam()).

Value

Predictions which the cross-validated model makes for xnew at the optimal value of lambda. Note that the default is the "lambda.1se" for lambda, to make this function consistent with cv.glmnet in the glmnet package.

The output depends on the ... argument which is passed on to the predict method for rgam objects.

Details

This function makes it easier to use the results of cross-validation to make a prediction.

Examples

Run this code

# NOT RUN {
set.seed(1)
n <- 100; p <- 20
x <- matrix(rnorm(n * p), n, p)
beta <- matrix(c(rep(2, 5), rep(0, 15)), ncol = 1)
y <- x %*% beta + rnorm(n)
cvfit <- cv.rgam(x, y)

# predictions at the lambda.1se value
predict(cvfit, xnew = x[1:5, ])

# predictions at the lambda.min value
predict(cvfit, xnew = x[1:5, ], s = "lambda.min")

# predictions at specific lambda value
predict(cvfit, xnew = x[1:5, ], s = 0.1)

# probability predictions for binomial family
bin_y <- ifelse(y > 0, 1, 0)
cvfit2 <- cv.rgam(x, bin_y, family = "binomial")
predict(cvfit2, xnew = x[1:5, ], type = "response", s = "lambda.min")

# }