根据k-近邻定义构建K近邻函数

利用knn定义构造函数，并通过十折交叉检验获取最优K值，最后计算算法精度（以iris数据为例）

1、利用k近邻定义，构造knn函数

knn_define <- function(train.data,test.data,k1){
  train.len = nrow(train.data)
  test.len = nrow(test.data)
  ## calculate distances between samples
  dist = as.matrix(dist(rbind(test.data[,1:4], train.data[,1:4]),
                        method= 'euclidean'))[1:test.len, (test.len+1):(test.len+train.len)]
  for (i in 1:dim(test.data)[1]) {
    nn = as.data.frame(sort(dist[i,], index.return = TRUE))[1:k1,2]
    test.label = train.data$Species[nn]
    labels = data.frame(sort(table(test.label),decreasing = T))[1,1]
    test.data[i,6] = labels
  }
  return(test.data)
}

2、利用十折交叉检验获取最优k

for (k1 in 3:15) {
  for (n in 1:10) {
    test.idx= seq(n,dim(iris)[1],by = 10)
    train.idx = setdiff(1:150,test.i