模型训练预测报错ValueError：特征名称与拟合时不匹配的解决方法

问题背景 #

训练模型时已通过删除InvoiceDate、BillingAddress等含字符串的列解决了字符串转浮点的报错，但执行预测时出现以下错误：

ValueError: The feature names should match those that were passed during fit. Feature names unseen at fit time:

相关代码片段：

X_train = X_train.drop(columns=['InvoiceDate', "BillingAddress", "BillingCity", "BillingState", "BillingCountry", "BillingPostalCode", "Rowversion_x", "Rowversion_y", "Rowversion", "Name", "Composer"], axis=1)

print(f'Tipo X_train: {type(X_train)} Tipo y_train: {type(y_train)}')
clf.fit(X_train, y_train)
y_pred = clf.predict(X_test)

accuracy = accuracy_score(y_test, y_pred)
print("Accuracy:", accuracy)

mining_data.to_csv('mining_table.csv', index=False)

解决办法 #

• 同步处理X_test的特征列
  你仅对X_train执行了列删除操作，但X_test保留了原始全部特征列，导致模型预测时的特征集与训练时不匹配。必须对X_test执行完全相同的列删除操作：

  # 用统一的列名列表处理，避免手动输入错误
  drop_cols = ['InvoiceDate', "BillingAddress", "BillingCity", "BillingState", "BillingCountry", "BillingPostalCode", "Rowversion_x", "Rowversion_y", "Rowversion", "Name", "Composer"]
  X_train = X_train.drop(columns=drop_cols, axis=1)
  X_test = X_test.drop(columns=drop_cols, axis=1)
  

• 验证特征列一致性
  在训练和预测前，检查X_train和X_test的特征列是否完全一致：

  # 打印特征列对比
  print("X_train特征列:", X_train.columns.tolist())
  print("X_test特征列:", X_test.columns.tolist())
  # 直接判断一致性
  print("特征列是否一致:", set(X_train.columns) == set(X_test.columns))
  

  若输出不一致，排查是否有其他代码修改了某一方的特征列。

• 用Pipeline封装预处理流程（可选）
  为避免后续重复手动处理特征列的问题，可使用sklearn.pipeline.Pipeline将预处理与模型训练封装在一起，确保训练和预测的预处理逻辑完全一致：

  from sklearn.pipeline import Pipeline
  from sklearn.base import BaseEstimator, TransformerMixin
  
  # 自定义删除列的变换器
  class DropColumns(BaseEstimator, TransformerMixin):
      def __init__(self, cols):
          self.cols = cols
      def fit(self, X, y=None):
          return self
      def transform(self, X):
          return X.drop(columns=self.cols, axis=1)
  
  # 构建Pipeline
  drop_cols = ['InvoiceDate', "BillingAddress", "BillingCity", "BillingState", "BillingCountry", "BillingPostalCode", "Rowversion_x", "Rowversion_y", "Rowversion", "Name", "Composer"]
  pipeline = Pipeline([
      ('drop_cols', DropColumns(cols=drop_cols)),
      ('classifier', clf)
  ])
  
  # 训练和预测
  pipeline.fit(X_train, y_train)
  y_pred = pipeline.predict(X_test)
  

内容的提问来源于stack exchange，提问作者information