retrieve intermediate features from a pipeline in Scikit (Python)

Question

I am using a pipeline very similar to the one given in this example :

>>> text_clf = Pipeline([('vect', CountVectorizer()),
...                      ('tfidf', TfidfTransformer()),
...                      ('clf', MultinomialNB()),
... ])

over which I use GridSearchCV to find the best estimators over a parameter grid.

However, I would like to get the column names of my training set with the get_feature_names() method from CountVectorizer(). Is this possible without implementing CountVectorizer() outside the pipeline?

Answer 1

Using the get_params() function, you can get access at the various parts of the pipeline and their respective internal parameters. Here's an example of accessing 'vect'

text_clf = Pipeline([('vect', CountVectorizer()),
                     ('tfidf', TfidfTransformer()),
                     ('clf', MultinomialNB())]
print text_clf.get_params()['vect']

yields (for me)

CountVectorizer(analyzer=u'word', binary=False, decode_error=u'strict',
    dtype=<type 'numpy.int64'>, encoding=u'utf-8', input=u'content',
    lowercase=True, max_df=1.0, max_features=None, min_df=1,
    ngram_range=(1, 1), preprocessor=None, stop_words=None,
    strip_accents=None, token_pattern=u'(?u)\\b\\w\\w+\\b',
    tokenizer=None, vocabulary=None)

I haven't fitted the pipeline to any data in this example, so calling get_feature_names() at this point will return an error.