我有一组字符串需要进行分类。我使用了一个管道对象。
我实现了两个虚拟转换器：一个将数据转换为特定格式（以便被另一个转换器接受），另一个将数据再次转换回其原始形式（一种逆转换）。
X和y是字符串列表，假设X=['London is great', 'London is beautiful', 'I hate London']和y=['p','p','n']。我希望X被转换为字符串的列表列表：X=[['London is great'], ['London is beautiful'], ['I hate London']]
我的代码如下：
from sklearn.naive_bayes import MultinomialNBfrom sklearn.feature_extraction.text import CountVectorizerfrom sklearn.feature_selection import SelectKBest, chi2from sklearn.pipeline import Pipelinefrom sklearn.base import TransformerMixin, BaseEstimatorvectorizer = CountVectorizer(input=u'content', analyzer=u'word', lowercase=True, stop_words=cached_stopwords, strip_accents=u'unicode', ngram_range=(1, 3), binary=False)estimators = [('pre_ds', PreprocessPreDS()), ('post_ds', PreprocesarPostDS()), ('vectorizer', vectorizer), ('feature_selector', SelectKBest(chi2, k=100)), ('clf', MultinomialNB())] # create the pipelinepipe = Pipeline(estimators)pipe.fit(X_train, y_train)其中我的自定义转换器如下：
class PreprocessPreDS(BaseEstimator, TransformerMixin): def __init__(self): pass def transform(self, X, *_): return [[x] for x in X] def fit(self, *_): return self def fit_transform(self, X, y=None, **fit_params): return self.fit(X) def get_params(self, deep=True): """ :param deep: ignored, as suggested by scikit learn's documentation :return: dict containing each parameter from the model as name and its current value """ return {} def set_params(self, **parameters): """ set all parameters for current objects :param parameters: dict containing its keys and values to be initialised :return: self """ for parameter, value in parameters.items(): setattr(self, parameter, value) return selfclass PreprocesarPostDS(BaseEstimator, TransformerMixin): def __init__(self): pass def transform(self, X, *_): return [x[0] for x in X] def fit(self, *_): return self def fit_transform(self, X, y=None, **fit_params): return self.fit(X) def get_params(self, deep=True): """ :param deep: ignored, as suggested by scikit learn's documentation :return: dict containing each parameter from the model as name and its current value """ return {} def set_params(self, **parameters): """ set all parameters for current objects :param parameters: dict containing its keys and values to be initialised :return: self """ for parameter, value in parameters.items(): setattr(self, parameter, value) return self当我运行这段代码时，我得到了以下错误：
Traceback (most recent call last): File "/home/rodrigo/nb/train_nb_pipeline.py", line 449, in <module> process(args.label, args.evaluate, args.label_all, corpus=args.corpus_path) File "/home/rodrigo/nb/train_nb_pipeline.py", line 179, in process pipe.fit(X_train, y_train) File "/home/rodrigo/.env/lib/python3.5/site-packages/sklearn/pipeline.py", line 248, in fit Xt, fit_params = self._fit(X, y, **fit_params) File "/home/rodrigo/.env/lib/python3.5/site-packages/sklearn/pipeline.py", line 213, in _fit **fit_params_steps[name]) File "/home/rodrigo/.env/lib/python3.5/site-packages/sklearn/externals/joblib/memory.py", line 362, in __call__ return self.func(*args, **kwargs) File "/home/rodrigo/.env/lib/python3.5/site-packages/sklearn/pipeline.py", line 581, in _fit_transform_one res = transformer.fit_transform(X, y, **fit_params) File "/home/rodrigo/.env/lib/python3.5/site-packages/sklearn/feature_extraction/text.py", line 869, in fit_transform self.fixed_vocabulary_) File "/home/rodrigo/.env/lib/python3.5/site-packages/sklearn/feature_extraction/text.py", line 790, in _count_vocab for doc in raw_documents:TypeError: 'PreprocessPostDS' object is not iterable然而，如果我从estimators中排除('pre_ds', PreprocessPreDS())和('post_ds', PreprocesarPostDS())，它就能正常运行。
回答：
将以下内容更改为：
def fit_transform(self, X, y=None, **fit_params): return self.fit(X)改为：
def fit_transform(self, X, y=None, **fit_params): return self.fit(X).transform(X)在上面的代码中，你实际上是返回self。self是类对象（在本例中是PreprocessPreDS和PreprocessPostDS）。fit_transform()应该返回转换后的数据，而不是类对象。
