NEWKNN.Rmd

---
title: "Untitled"
output: html_document
---






```{r}
datainput=function(max,goal){
  data=c()
  index1=c()
  index2=c()
    for (i in 0:9){
      aa=paste(goal,i,"_",0:num[i+1],".txt",sep="")
      bb=rep(i,num[i+1]+1)
      cc=0:num[i+1]
      index1=c(index1,bb)
      index2=c(index2,cc)
      data=c(data,aa)
    }
#############製作所有目錄，並記錄檔案編號到index1 ,index2
  xx=matrix(0,ncol=32*32+2,nrow=length(data))
    for (i in 1:length(data)){
      x=readLines(data[i])
      x1=strsplit(x,split="")
      x1=unlist(x1)
      x1=as.numeric(x1)
      xx[i,3:1026]=x1
    }
  xx[,1]=index1
  xx[,2]=index2
  return(xx)
}
##############xx 為第一行是編號0~9 第二行為第二個編號  第三行到第1026行是將該個資料的data 以數值存入


#########讀入資料
num=c(188,197,194,198,185,186,194,200,179,203)
train=datainput(max=num,goal="C:\\Users\\tsr\\Desktop\\程式\\trainingDigits\\")
num=c(86,96,91,84,113,107,86,95,90,88)
test=datainput(max=num,goal="C:\\Users\\tsr\\Desktop\\程式\\testDigits\\")
#################判斷開始
train=as.data.frame(train)
test=as.data.frame(test)
class(test)
count=0
k=3
library(FNN)
for (i in 1:dim(test)[1]){
    aa=get.knnx(data=train[,3:1026],query=test[i,3:1026],k=k,"kd_tree")
    pre=train[aa$nn.index,1]
    gggg=table(pre)
    p=names(gggg)[which.max(gggg)]
    if  (p==test[i,1]) count=count+1
    }
count

per=count/dim(test)[1]
sprintf("判斷成功率為%s",per)

```