ValueError: Input array dimensions not right for CountVectorizer()

Question

在 sklearn 管道中使用 make_column_transformer() 时，我在尝试使用 CountVectorizer 时遇到错误。

我的 DataFrame 有两列，'desc-title' 和 'SPchangeHigh'。这是两行的片段：

features = pd.DataFrame([["T. Rowe Price sells most of its Tesla shares", .002152],
                         ["Gannett to retain all seats in MNG proxy fight", 0.002152]],
                        columns=["desc-title", "SPchangeHigh"])

我能够运行以下管道没有问题：

preprocess = make_column_transformer(
    (StandardScaler(),['SPchangeHigh']),
    ( OneHotEncoder(),['desc-title'])
)
preprocess.fit_transform(features.head(2))

然而，当我用 CountVectorizer(tokenizer=tokenize) 替换 OneHotEncoder() 时，它失败了：

preprocess = make_column_transformer(
    (StandardScaler(),['SPchangeHigh']),
    ( CountVectorizer(tokenizer=tokenize),['desc-title'])
)
preprocess.fit_transform(features.head(2))

我得到的错误是：

ValueError                                Traceback (most recent call last)
<ipython-input-71-d77f136b9586> in <module>()
      3     ( CountVectorizer(tokenizer=tokenize),['desc-title'])
      4 )
----> 5 preprocess.fit_transform(features.head(2))

C:\anaconda3\lib\site-packages\sklearn\compose\_column_transformer.py in fit_transform(self, X, y)
    488         self._validate_output(Xs)
    489 
--> 490         return self._hstack(list(Xs))
    491 
    492     def transform(self, X):

C:\anaconda3\lib\site-packages\sklearn\compose\_column_transformer.py in _hstack(self, Xs)
    545         else:
    546             Xs = [f.toarray() if sparse.issparse(f) else f for f in Xs]
--> 547             return np.hstack(Xs)
    548 
    549 

C:\anaconda3\lib\site-packages\numpy\core\shape_base.py in hstack(tup)
    338         return _nx.concatenate(arrs, 0)
    339     else:
--> 340         return _nx.concatenate(arrs, 1)
    341 
    342 

ValueError: all the input array dimensions except for the concatenation axis must match exactly

如果有人能帮助我，我将不胜感激。

Answer 1

删除 'desc-title' 周围的括号。您需要一个 one-dimensional 数组，而不是列向量。

preprocess = make_column_transformer(
    (StandardScaler(),['SPchangeHigh']),
    ( CountVectorizer(),'desc-title')
)
preprocess.fit_transform(features.head(2))

Sklearn documentation describes this nuanced specification:

The difference between specifying the column selector as 'column' (as a simple string) and ['column'] (as a list with one element) is the shape of the array that is passed to the transformer. In the first case, a one dimensional array will be passed, while in the second case it will be a 2-dimensional array with one column, i.e. a column vector

...

Be aware that some transformers expect a 1-dimensional input (the label-oriented ones) while some others, like OneHotEncoder or Imputer, expect 2-dimensional input, with the shape [n_samples, n_features].

ValueError: Input array dimensions not right for CountVectorizer()

ValueError: Input array dimensions not right for CountVectorizer()

python

pipeline

scikit-learn