flars (version 1.0)

predict.flars: Prediction for functional least angle regression.

Description

This is the function that carries out the prediction of the new observations.

Usage

"predict"(object,newdata,...)

Arguments

object

This must be a flars object from the function flars.

newdata

A list of new observations. The format of this set of data must be the same as the training data, including the order of the variables.

...

Other arguments to input.

Value

A matrix of predictions. Since the input flars object may have more than one estimated coefficients, the number of predictions may be more than one set. Each column of the outcome is corresponding to one set of coefficients.

Examples

library(flars)
library(fda)
## Generate some data.
dataL=data_generation(seed = 1,uncorr = TRUE,nVar = 8,nsamples = 120,
      var_type = 'm',cor_type = 3)

## Split the training data and the testing data
nTrain=80
nsamples=120

TrainIdx=seq(nTrain)
TestIdx=seq(nsamples)[-TrainIdx]
fsmTrain=lapply(dataL$x,function(fsmI) fsmI[TrainIdx,,drop=FALSE])
fsmTest=lapply(dataL$x,function(fsmI) fsmI[TestIdx,,drop=FALSE])
yTrain=dataL$y[TrainIdx]
yTest=dataL$y[TestIdx]
  
## Do the variable selection
out=flars(fsmTrain,yTrain,method='basis',max_selection=9,
    normalize='norm',lasso=FALSE)
	
## Do the prediction
pred=predict(out,newdata = fsmTest)

# apply(pred,2,flars:::rmse,yTest)