knn.reg: k Nearest Neighbor Regression

In FNN: Fast Nearest Neighbor Search Algorithms and Applications

k Nearest Neighbor Regression

Description

k-nearest neighbor regression

Usage

knn.reg(train, test = NULL, y, k = 3, algorithm=c("kd_tree", 
        "cover_tree", "brute"))

Arguments

train	matrix or data frame of training set cases.
test	matrix or data frame of test set cases. A vector will be interpreted as a row vector for a single case. If not supplied, cross-validataion will be done.
y	reponse of each observation in the training set.
k	number of neighbours considered.
algorithm	nearest neighbor search algorithm.

Details

If test is not supplied, Leave one out cross-validation is performed and R-square is the predicted R-square.

Value

knn.reg returns an object of class "knnReg" or "knnRegCV" if test data is not supplied.

The returnedobject is a list containing at least the following components:

call	the match call.
k	number of neighbours considered.
n	number of predicted values, either equals test size or train size.
pred	a vector of predicted values.
residuals	predicted residuals. NULL if test is supplied.
PRESS	the sums of squares of the predicted residuals. NULL if test is supplied.
R2Pred	predicted R-square. NULL if test is supplied.

Note

The code for “VR” nearest neighbor searching is taken from class source

Author(s)

Shengqiao Li. To report any bugs or suggestions please email: [e-mail]

See Also

knn.

Examples

  if(require(chemometrics)){
    data(PAC);
    pac.knn<- knn.reg(PAC$X, y=PAC$y, k=3);
    
    plot(PAC$y, pac.knn$pred, xlab="y", ylab=expression(hat(y)))
  } 

FNN documentation built on Sept. 22, 2024, 9:06 a.m.