导入仅适用于 Python 函数

Question

背景信息：我正在使用 scikit-learn 开发模型。我使用 sklearn.cross_validation 模块将数据分成单独的训练和测试集，如下所示：

def train_test_split(input_data):
       from sklearn.cross_validation import train_test_split

        ### STEP 1: Separate y variable and remove from X
       y = input_data['price']
       X = input_data.copy()
       X.drop('price', axis=1, inplace=True)

        ### STEP 2: Split into training & test sets
       X_train, X_test, y_train, y_test =\ 
                        train_test_split(X, y, test_size=0.2, random_state=0)
       return X_train, X_test, y_train, y_test

我的问题：当我尝试在我的函数之外导入 sklearn.cross_validation 模块时，像这样，我收到以下错误：

from sklearn.cross_validation import train_test_split

def train_test_split(input_data):
       ### STEP 1: Separate y variable and remove from X
       y = input_data['price']
       X = input_data.copy()
       X.drop('price', axis=1, inplace=True)

       ### STEP 2: Split into training & test sets
       X_train, X_test, y_train, y_test =\ 
                        train_test_split(X, y, test_size=0.2, random_state=0)
       return X_train, X_test, y_train, y_test

错误：

TypeError: train_test_split() got an unexpected keyword argument 'test_size'

知道为什么吗？

Answer 1

您正在从 sklear.cross_validation 导入函数 train_test_split，然后用本地函数 train_test_split 覆盖名称。

试试：

from sklearn.cross_validation import train_test_split as sk_train_test_split

def train_test_split(input_data):
       ### STEP 1: Separate y variable and remove from X
       y = input_data['price']
       X = input_data.copy()
       X.drop('price', axis=1, inplace=True)

       ### STEP 2: Split into training & test sets
       X_train, X_test, y_train, y_test =\ 
                        sk_train_test_split(X, y, test_size=0.2, random_state=0)  # use the imported function instead of local one
       return X_train, X_test, y_train, y_test