python中的KNN和交叉验证实现

如何解决python中的KNN和交叉验证实现

我正在努力尝试实施 K-NN 方法 交叉验证。

我必须找到 k 的最优值，代表最优超参数 邻居。数据集由 10 折（训练数据）组成，每个 尺寸为 [99x64]； 10 个训练标签 大小（99x1），及其对应的测试数据 (110x1) 和 (110x64)。

计算欧几里得距离后应该遵循哪些步骤？

出于测试目的，我只设置了 3 折，这给了我 6 个距离向量。 这是我的代码：

# Assigning the parameters
K = 1 # Total Neighbours
folds = 10 # Total Folds
lambda_param = 0.1
dataf = [f for f in train_data if isfile(join(root_dir,f))]
datal = [f for f in train_label if isfile(join(root_dir,f))]

def F_train_sets(F,data):
  return [pd.DataFrame.to_numpy(pd.read_csv(join(root_dir,f))) for f in data[:F] ]
total_trainf = F_train_sets(folds,dataf) #Total train set for X [99x64x10]
total_trainl = F_train_sets(folds,datal) #Total label set for Y [99x1]

def train_fn(k,fold,data,labels):
  train_data = data
  validation_dataset = data[fold-1]
  training_set = []
  accum = None
  i = 0
  while i < len(data):
    if(i== fold):
      pass
    else:
      training_set.append(data[i])
    i+=1
  for ts in training_set:
    dist = (validation_dataset - ts)**2
    dist = np.sum(dist,axis=1)
    dist = np.sqrt(dist)

i=0
while i < K:
  f=1
  while f <= folds:
    train_fn(i,f,total_trainf,total_trainl[0])
    f+=1
  i+=1