predict {crisp}

Predicts Observations for a New Covariate Matrix using Fit from crisp or crispCV.

Description

This function makes predictions for a specified covariate matrix for a fit of the class crispCV, or class crisp with a user-specified tuning parameter.

Usage

## S3 method for class 'crisp'
predict(object, new.X, lambda.index, ...)

## S3 method for class 'crispCV'
predict(object, new.X, ...)

Arguments

object

An object of class crisp or crispCV, which result from running the crisp or crispCV functions, respectively.

new.X

The covariate matrix for which to make predictions.

lambda.index

The index for the desired value of lambda, i.e., object$lambda.seq[lambda.index].

...

Additional arguments to be passed, which are ignored in this function.

Details

The ith prediction is made to be the value of object$M.hat.list[[lambda.index]] corresponding to the pair of covariates closest (in Euclidean distance) to new.X[i,].

Value

A vector containing the fitted y values for new.X.

Examples

## Not run: 
#See ?'crisp-package' for a full example of how to use this package

#generate data (using a very small 'n' for illustration purposes)
set.seed(1)
data <- sim.data(n = 15, scenario = 2)

#fit model for a range of tuning parameters, i.e., lambda values
#lambda sequence is chosen automatically if not specified
crisp.out <- crisp(X = data$X, y = data$y)
#or fit model and select lambda using 2-fold cross-validation
#note: use larger 'n.fold' (e.g., 10) in practice
crispCV.out <- crispCV(X = data$X, y = data$y, n.fold = 2)

#we can make predictions for a covariate matrix with new observations
#new.X with 20 observations
new.data <- sim.data(n = 20, scenario = 2)
new.X <- new.data$X
#these will give the same predictions:
yhat1 <- predict(crisp.out, new.X = new.X, lambda.index = crispCV.out$index.cv)
yhat2 <- predict(crispCV.out, new.X = new.X)

## End(Not run)