Python - NLTK 训练/测试拆分

Question

我一直在关注 SentDex 的 video series 关于 NLTK 和 Python，并构建了一个脚本，该脚本使用各种模型来确定评论情绪，例如逻辑回归。我担心的是，我认为 SentDex 的方法在确定用于训练的单词时包括测试集，这显然是不可取的（训练/测试拆分发生在特征选择之后）。

（根据 Mohammed Kashif 的 cmets 编辑）

完整代码：

import nltk
import numpy as np
from nltk.classify.scikitlearn import SklearnClassifier
from nltk.classify import ClassifierI
from nltk.corpus import movie_reviews
from sklearn.naive_bayes import MultinomialNB

documents = [ (list(movie_reviews.words(fileid)), category)
             for category in movie_reviews.categories()
             for fileid in movie_reviews.fileids(category) ]

all_words = []

for w in movie_reviews.words():
    all_words.append(w.lower())

all_words = nltk.FreqDist(all_words)

word_features = list(all_words.keys())[:3000]

def find_features(documents):
    words = set(documents)
    features = {}
    for w in word_features:
        features[w] = (w in words)

    return features

featuresets = [(find_features(rev), category) for (rev, category) in documents]

np.random.shuffle(featuresets)

training_set = featuresets[:1800]
testing_set = featuresets[1800:]

MNB_classifier = SklearnClassifier(MultinomialNB())
MNB_classifier.train(training_set)
print("MNB_classifier accuracy:", (nltk.classify.accuracy(MNB_classifier, testing_set)) *100)

已经试过了：

documents = [ (list(movie_reviews.words(fileid)), category)
             for category in movie_reviews.categories()
             for fileid in movie_reviews.fileids(category) ]

np.random.shuffle(documents)

training_set = documents[:1800]
testing_set = documents[1800:]

all_words = []

for w in documents.words():
    all_words.append(w.lower())

all_words = nltk.FreqDist(all_words)

word_features = list(all_words.keys())[:3000]

def find_features(training_set):
    words = set(training_set)
    features = {}
    for w in word_features:
        features[w] = (w in words)

    return features

featuresets = [(find_features(rev), category) for (rev, category) in training_set]

np.random.shuffle(featuresets)

training_set = featuresets
testing_set = testing_set

MNB_classifier = SklearnClassifier(MultinomialNB())
MNB_classifier.train(training_set)
print("MNB_classifier accuracy:", (nltk.classify.accuracy(MNB_classifier, testing_set)) *100)

产生错误：

Traceback（最近一次调用最后一次）：

文件“”，第 34 行，在 print("MNB_classifier 精度：", (nltk.classify.accuracy(MNB_classifier, testing_set)) *100)

文件“C:\ProgramData\Anaconda3\lib\site-packages\nltk\classify\util.py”，第 87 行，准确无误 results = classifier.classify_many([fs for (fs, l) in gold])

文件“C:\ProgramData\Anaconda3\lib\site-packages\nltk\classify\scikitlearn.py”，第 85 行，在分类多 X = self._vectorizer.transform(featuresets)

文件“C:\ProgramData\Anaconda3\lib\site-packages\sklearn\feature_extraction\dict_vectorizer.py”，第 291 行，在转换中 return self._transform(X,fitting=False)

文件“C:\ProgramData\Anaconda3\lib\site-packages\sklearn\feature_extraction\dict_vectorizer.py”，第 166 行，在 _transform for f, v in Six.iteritems(x):

文件“C:\ProgramData\Anaconda3\lib\site-packages\sklearn\externals\six.py”，第 439 行，在 iteritems 返回 iter(getattr(d, _iteritems)(**kw))

AttributeError: 'list' 对象没有属性 'items'

Answer 1

好的，所以代码中有几个错误。我们将一一进行。

首先，您的documents 列表是一个元组列表，它没有words() 方法。为了访问所有单词，请像这样更改 for 循环

all_words = []

for words_list, categ in documents:      #<-- each wordlist is a list of words
    for w in words_list:                 #<-- Then access each word in list
        all_words.append(w.lower())

其次，您需要为 training 和 test 集创建功能集。您只使用了training_set 的功能集。把代码改成这样

featuresets = [(find_features(rev), category) for (rev, category) in documents]

np.random.shuffle(featuresets)

training_set = featuresets[:1800]
testing_set = featuresets[1800:]

所以最终的代码就变成了

documents = [ (list(movie_reviews.words(fileid)), category)
             for category in movie_reviews.categories()
             for fileid in movie_reviews.fileids(category) ]

np.random.shuffle(documents)

training_set = documents[:1800]
testing_set = documents[1800:]

all_words = []

for words_list, categ in documents:
    for w in words_list:
        all_words.append(w.lower())

all_words = nltk.FreqDist(all_words)

word_features = list(all_words.keys())[:3000]

def find_features(training_set):
    words = set(training_set)
    features = {}
    for w in word_features:
        features[w] = (w in words)

    return features

featuresets = [(find_features(rev), category) for (rev, category) in documents]

np.random.shuffle(featuresets)

training_set = featuresets[:1800]
testing_set = featuresets[1800:]

MNB_classifier = SklearnClassifier(MultinomialNB())
MNB_classifier.train(training_set)
print("MNB_classifier accuracy:", (nltk.classify.accuracy(MNB_classifier, testing_set)) *100)