ColumnTransformer 在管道中失败并显示 CountVectorizer/HashingVectorizer（多个文本特征）

Question

与此问题类似 () 我想在管道中使用 ColumnTransformer 在具有文本功能的列上应用 CountVectorizer/HashingVectorizer。但是我没有只有一个文本功能，而是多个。如果我传递单个功能（不是列表，就像另一个问题的解决方案中所建议的那样）它工作正常，我该如何为多个做？

numeric_features = ['x0', 'x1', 'y0', 'y1']
categorical_features = []
text_features = ['text_feature', 'another_text_feature']

numeric_transformer = Pipeline(steps=[('scaler', StandardScaler())])
categorical_transformer = Pipeline(steps=[('encoder', OneHotEncoder())])
text_transformer = Pipeline(steps=[('hashing', HashingVectorizer())])

preprocessor = ColumnTransformer(transformers=[
    ('numeric', numeric_transformer, numeric_features), 
    ('categorical', categorical_transformer, categorical_features),
    ('text', text_transformer, text_features)
])
    
steps = [('preprocessor', preprocessor),
         ('clf', SGDClassifier())]
    
pipeline = Pipeline(steps=steps)
    
pipeline.fit(X_train, y_train)

Answer 1

只需为每个文本特征使用一个单独的转换器。

preprocessor = ColumnTransformer(transformers=[
    ('numeric', numeric_transformer, numeric_features), 
    ('categorical', categorical_transformer, categorical_features),
    ('text', text_transformer, 'text_feature'),
    ('more_text', text_transformer, 'another_text_feature'),
])

（变压器在拟合过程中被克隆，所以你将有两个单独的 text_transformer 副本，一切都很好。如果你担心像这样指定同一个变压器两次，你总是可以 copy/clone 它在指定 ColumnTransformer 之前手动设置。)

ColumnTransformer 在管道中失败并显示 CountVectorizer/HashingVectorizer（多个文本特征）

ColumnTransformer fails with CountVectorizer/HashingVectorizer in a pipeline (multiple textfeatures)

python

scikit-learn

pipeline