predict.stepplr {stepPlr} | R Documentation |
This function computes the linear predictors, probability estimates,
or the class labels for new data, using a stepplr
object.
predict.stepplr(object, x = NULL, newx = NULL, type = c("link", "response", "class"), ...)
object |
a stepplr object
|
x |
the matrix of features used for fitting object. If
newx is provided, x must be provided as well.
|
newx |
a matrix of features at which the predictions are made. If
newx=NULL, predictions for the training data are returned.
|
type |
If type=link, the linear predictors are returned; if
type=response, the probability estimates are returned; and if
type=class, the class labels are returned. Default is
type=link.
|
... |
other options for the prediction |
Mee Young Park and Trevor Hastie
Mee Young Park and Trevor Hastie (2006) Penalized Logistic Regression for Detecting Gene Interactions - available at the authors' websites, http://stat.stanford.edu/~mypark or http://stat.stanford.edu/~hastie/pub.htm.
stepplr
n <- 100 p <- 5 x0 <- matrix(sample(seq(3),n*p,replace=TRUE),nrow=n) x0 <- cbind(rnorm(n),x0) y <- sample(c(0,1),n,replace=TRUE) level <- vector("list",length=6) for (i in 2:6) level[[i]] <- seq(3) fit <- step.plr(x0,y,level=level) x1 <- matrix(sample(seq(3),n*p,replace=TRUE),nrow=n) x1 <- cbind(rnorm(n),x1) pred1 <- predict(fit,x0,x1,type="link") pred2 <- predict(fit,x0,x1,type="response") pred3 <- predict(fit,x0,x1,type="class")