Learn R Programming

l2boost (version 1.0.3)

predict.l2boost: predict method for l2boost models.

Description

predict is a generic function for predictions from the results of various model fitting functions.

@details predict.l2boost takes the optional xnew (equivalent newdata) data.frame and returns the model estimates from an l2boost object. If neither xnew or newdata are provided, predict returns estimates for the l2boost training data set.

By default, predict.l2boost returns the function estimates, unless type="coef" then the set of regression coefficients (beta) are returned from the l2boost object.

Usage

# S3 method for l2boost
predict(object, xnew = NULL, type = c("fit", "coef"), newdata = xnew, ...)

Arguments

object

an l2boost object

xnew

a new design matrix to fit with the l2boost object

type

"fit" or "coef" determins the values returned. "fit" returns model estimates, "coef" returns the model coefficients

newdata

a new design matrix to fit with the l2boost object

...

other arguments (currently not used)

Value

function estimates for type=fit, coefficient estimates for type=coef

  • yhatvector of n function estimates from the final step M

  • yhat.pathlist of M function estimates, one at each step m

or

  • coefvector of p beta coefficient estimates from final step M

  • coef.standvector of p standardized beta coefficient estimates from final step M

  • coef.pathlist of vectors of p beta coefficient estimates, one for each step m

  • coef.stand.pathlist of vectors of p standardized beta coefficient estimates, one for each step m

See Also

predict and l2boost, coef.l2boost, fitted.l2boost, residuals.l2boost and cv.l2boost

Examples

Run this code
# NOT RUN {
#--------------------------------------------------------------------------
# Example 1: Diabetes 
#  
# See Efron B., Hastie T., Johnstone I., and Tibshirani R. 
# Least angle regression. Ann. Statist., 32:407-499, 2004.
data(diabetes)

object <- l2boost(diabetes$x,diabetes$y, M=1000, nu=.01)

# With no arguments returns the estimates at the full M from the training data.
prd <- predict(object)
prd$yhat

# at step m=600
prd$yhat.path[[600]]

# Also can return coefficient estimates. This is equivalent to \code{\link{coef.l2boost}}
cf <- predict(object, type="coef")
cf$coef

# at step m=600
cf$coef.path[[600]]

# Or used to predict new data, in this case a subset of training data
cbind(diabetes$y[1:5], predict(object, xnew=diabetes$x[1:5,])$yhat)

# }

Run the code above in your browser using DataLab