关于使用R的KNN的k重交叉验证的问题

library(class)
library(caret)

# 示例数据
data(iris)
iris <- iris[,-5]  # 去掉类别标签列

# K重交叉验证选择最佳K值
k_values <- 1:20
accuracy <- numeric(length(k_values))

for (k in k_values) {
  folds <- createFolds(iris[,5], k = 10)
  acc <- numeric(length(folds))
  for (i in 1:length(folds)) {
    train_data <- iris[-folds[[i]],]
    test_data <- iris[folds[[i]],]
    pred <- knn(train = train_data[,1:4], test = test_data[,1:4], cl = train_data[,5], k = k)
    acc[i] <- sum(pred == test_data[,5]) / length(pred)
  }
  accuracy[k] <- mean(acc)
}

# 找到最佳K值
best_k <- k_values[which.max(accuracy)]
print(best_k)

# 特征缩放
iris_scaled <- scale(iris[,1:4])

# 使用缩放后的数据进行KNN分类
pred <- knn(train = iris_scaled[,-5], test = iris_scaled[,-5], cl = iris[,5], k = best_k)

# 示例：使用SMOTE进行过采样
library(DMwR)

iris_imbalanced <- iris
iris_imbalanced[,5] <- ifelse(runif(nrow(iris)) < 0.8, "setosa", "versicolor")

iris_resampled <- SMOTE(class ~ ., data = iris_imbalanced, perc.over = 100, perc.under = 200)
pred_resampled <- knn(train = iris_resampled[,1:4], test = iris_resampled[,1:4], cl = iris_resampled[,5], k = best_k)