我正在尝试将我的文本分类模型进行 pickle 处理，并重新加载到 Flask 应用程序界面中。

我有一个特定的函数作为分析器使用，称为 split_into_lemmas

def split_into_lemmas(message):    message = unicode(message, 'utf8').lower()    words = TextBlob(message).words    # 对每个单词，取其“基础形式” = 词形还原     return [word.lemma for word in words]from sklearn.pipeline import Pipelinecount_vect = CountVectorizer(analyzer=split_into_lemmas,ngram_range= (1, 3), encoding='utf8',stop_words =None)tfidf_transformer = TfidfTransformer()text_clf = Pipeline([('vect', count_vect), ('tdif', tfidf_transformer), ('clf', best_svc)])%%timetext_clf.fit(X=data['Condition'], y=data['condition_predict'])

我训练了模型并通过 pickle 保存它

_ = joblib.dump(text_clf, 'classification_pipeline.pkl')

另一方面
当我尝试重新加载管道时

我得到了以下错误

---------------------------------------------------------------------------AttributeError                            Traceback (most recent call last)<ipython-input-3-bb0859b3946a> in <module>()      6       7 clf_pipeline = open('C:/Users/Falco/Desktop/directory/WRMD_paper/classification_pipeline.pkl','rb')----> 8 clf = joblib.load(clf_pipeline)C:\ProgramData\Anaconda2\lib\site-packages\sklearn\externals\joblib\numpy_pickle.pyc in load(filename, mmap_mode)    586         filename = getattr(fobj, 'name', '')    587         with _read_fileobject(fobj, filename, mmap_mode) as fobj:--> 588             obj = _unpickle(fobj)    589     else:    590         with open(filename, 'rb') as f:C:\ProgramData\Anaconda2\lib\site-packages\sklearn\externals\joblib\numpy_pickle.pyc in _unpickle(fobj, filename, mmap_mode)    524     obj = None    525     try:--> 526         obj = unpickler.load()    527         if unpickler.compat_mode:    528             warnings.warn("The file '%s' has been generated with a "C:\ProgramData\Anaconda2\lib\pickle.pyc in load(self)    862             while 1:    863                 key = read(1)--> 864                 dispatch[key](self)    865         except _Stop, stopinst:    866             return stopinst.valueC:\ProgramData\Anaconda2\lib\pickle.pyc in load_global(self)   1094         module = self.readline()[:-1]   1095         name = self.readline()[:-1]-> 1096         klass = self.find_class(module, name)   1097         self.append(klass)   1098     dispatch[GLOBAL] = load_globalC:\ProgramData\Anaconda2\lib\pickle.pyc in find_class(self, module, name)   1130         __import__(module)   1131         mod = sys.modules[module]-> 1132         klass = getattr(mod, name)   1133         return klass   1134 AttributeError: 'module' object has no attribute 'split_into_lemmas'

当我在笔记本中重新声明该函数时，模型加载良好并运行，但当我将笔记本保存为 .py 文件并作为 Flask 应用程序运行时，它不运行并给我相同的错误。

有人能帮助我正确保存管道，这样我就不必声明该函数了吗？

回答：

当你重新加载 pickle 时，你也需要定义 split_into_lemmas 函数。