KNNperfor<-function(trainX, trainy, testX, testy,k)

{##build KNN classification model

## example of euclidean distances.

M=length(trainX[,1])

Xt=rbind(trainX, testX)

Dtest=as.matrix(dist(Xt, method = "euclidean"))

Tsort=sort(Dtest[M+1,1:M], method = "qu", index.return = TRUE)

Vote=trainy[Tsort$ix[1:k]]

PreY=as.numeric(names(table(Vote))[table(Vote)==max(table(Vote))])  

Error=PreY!=testy

Predict=c(PreY,Error)

return(Predict)

}

KNN分类时确定k后得到投票向量Vote，用以下可求出预测分类

PreY=as.numeric(names(table(Vote))[table(Vote)==max(table(Vote))])