Scikit learn 为什么n-gram范围去掉了中性标记和概率？

为什么上面缺少9%:Neutral？

你可以发布一个y值的摘要（比如标签和每个标签的样本数）？'13%'：'sazing'，'15%'：'Bad'，'57%'：'Good'，'6%'：'sorry'：通常是9%。不仅是这批的结果。

from sklearn.feature_extraction.text import CountVectorizer

vectorizer = CountVectorizer(stop_words = "english", ngram_range=(1, 2), max_features=5000)
X = vectorizer.fit_transform(data['reviews_text'])
bag_of_words= pd.DataFrame(X.toarray(), columns=vectorizer.get_feature_names())
x= bag_of_words
y= data['reviews_sentiment']

from sklearn.naive_bayes import MultinomialNB
model = MultinomialNB()
trained_model = model.fit(X, y)
probas= trained_model.predict_proba(x_test)[0]
probabilities=[str(int(x*100))+'%' for x in probas]
labels= list(trained_model.classes_)

dict(zip(probabilities,labels))

{'13%': 'Amazing', '15%': 'Bad', '57%': 'Good', '6%': 'Terrible'}