Package 'kNNvs'

Title: k Nearest Neighbors with Grid Search Variable Selection
Description: k Nearest Neighbors with variable selection, combine grid search and forward selection to achieve variable selection in order to improve k Nearest Neighbors predictive performance.
Authors: Eddie Pei [aut, cre], Ernest Fokoue [aut]
Maintainer: Eddie Pei <[email protected]>
License: GPL-3
Version: 0.1.0
Built: 2025-02-13 03:25:37 UTC
Source: https://github.com/cran/kNNvs

Help Index

k Nearest Neighbors with Grid Search Variable Selection

Description

k Nearest Neighbors with Grid Search Variable Selection

Usage

kNNvs(
  train_x,
  test_x,
  cl_train,
  cl_test,
  k,
  model = c("regression", "classifiation")
)

Arguments

train_x

matrix or data frame of training set
test_x

matrix or data frame of test set
cl_train

factor of true classifications of training set
cl_test

factor of true classifications of test set
k

the number of neighbors
model

regression or classifiation

Details

kNNvs is simply use add one and then compare acc to pick the best variable set for the knn model

Value

ACC or MSE, best variable combination, estimate value yhat

Examples

{
   data(iris3)
   train_x <- rbind(iris3[1:25,,1], iris3[1:25,,2], iris3[1:25,,3])
   test_x <- rbind(iris3[26:50,,1], iris3[26:50,,2], iris3[26:50,,3])
   cl_train<- cl_test<- factor(c(rep("s",25), rep("c",25), rep("v",25)))
   k<- 5
   # cl_test is not null
   mymodel<-kNNvs(train_x,test_x,cl_train,cl_test,k,model="classifiation")
   mymodel
   # cl_test is null
   mymodel<-kNNvs(train_x,test_x,cl_train,cl_test=NULL,k,model="classifiation")
   mymodel
   }