sklearn.pipeline 如何手动工作？

Question

Currently, I am working on the sklearn.pipeline which is just wonderful Here is an example:

model = make_pipeline(TfidfVectorizer(), MultinomialNB())
model.fit(train.data, train.target)
labels = model.predict(test.data)

(*data is from train = fetch_20newsgroups(subset='train', categories=categories )) with categories= ['talk.religion.misc', 'soc.religion.christian', 'sci.space','comp.graphics']

However, my understanding is just still very vague. I would like to ask that if we do it step by step without pipeline how it could be. Here is just what I am trying to do but it failed.

from sklearn.datasets import fetch_20newsgroups
Categories = ['talk.religion.misc', 'soc.religion.christian', 'sci.space','comp.graphics']
train = fetch_20newsgroups(subset='train', categories=categories)`

from sklearn.feature_extraction.text import TfidfVectorizer
model1=TfidfVectorizer()
X=model1.fit_transform(train.data)

from sklearn.naive_bayes import MultinomialNB
model2=MultinomialNB
model2.fit(....)

At this far, I just don't know what to do next because the shape of X is not suitable for model2 .

For your further information of this, go to the book from this link at page (406/548)

*** Please pardon for my silly question. I know I can do it by using pipeline but just want to try

Answer 1

You are almost there! you need to use MultinomialNB() instead of MultinomialNB .

Try the following procedure.

from sklearn.datasets import fetch_20newsgroups
Categories = ['talk.religion.misc', 'soc.religion.christian', 'sci.space','comp.graphics']
train = fetch_20newsgroups(subset='train', categories=categories)


from sklearn.feature_extraction.text import TfidfVectorizer
model1=TfidfVectorizer()
X=model1.fit_transform(train.data)

from sklearn.naive_bayes import MultinomialNB
model2=MultinomialNB()
model2.fit(X, train.target)
model2.predict(model1.transform(test.data))

# array([2, 1, 1, ..., 2, 1, 1])