OneHotEncoder对于pipeline中的分类变量不会自动删除。

可以使用ColumnTransformer来为pipeline定义特定的数据预处理步骤，并使用OneHotEncoder作为分类变量的处理方式。这样可以确保分类变量被正确预处理并删除原始变量。以下是一个示例代码：

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import OneHotEncoder
from sklearn.pipeline import Pipeline
from sklearn.impute import SimpleImputer
from sklearn.linear_model import LogisticRegression

# 定义数据预处理步骤
preprocessing = ColumnTransformer(
    transformers=[
        ('num', SimpleImputer(strategy='median'), ['age', 'fare']),
        ('cat', OneHotEncoder(), ['sex', 'embarked'])
    ],
    # 移除原始分类变量
    remainder='drop'
)

# 定义pipeline
pipeline = Pipeline([
    ('preprocess', preprocessing),
    ('model', LogisticRegression())
])

# 训练模型
pipeline.fit(X_train, y_train)

# 预测
y_pred = pipeline.predict(X_test)

# 评估
accuracy = pipeline.score(X_test, y_test)