Question

我在np.array中有5个句子，我想找到最常见的n个出现的单词及其相对计数。例如，如果n为3，我希望使用3个最常用的单词。作为相对计数，我希望将单词出现的次数除以单词总数。我在下面有一个例子：

0    oh i am she cool though might off her a brownie lol
1    so trash wouldnt do colors better tweet
2    love monkey brownie as much as a tweet
3    monkey get this tweet around i think
4    saw a brownie to make me some monkey

在上一个问题的帮助下，我设法找到了最常用的单词

import numpy as np
from sklearn.feature_extraction.text import CountVectorizer

A = np.array(["oh i am she cool though might off her a brownie lol", 
              "so trash wouldnt do colors better tweet", 
              "love monkey brownie as much as a tweet",
              "monkey get this tweet around i think",
              "saw a brownie to make me some monkey" ])

n = 3
vectorizer = CountVectorizer()
X = vectorizer.fit_transform(A)

vocabulary = vectorizer.get_feature_names()
ind  = np.argsort(X.toarray().sum(axis=0))[-n:]

top_n_words = [vocabulary[a] for a in ind]

print (top_n_words)
['tweet', 'monkey', 'brownie']

但是，现在我想找到相对计数？有没有简单的pythonic方法来做到这一点？例如：

print (top_n_words_relative_count)
[3/42, 3/42, 3/42]

其中42是总字数。

Answer 1

您可以使用collections.Counter：

>>> A = np.array(["oh i am she cool though might off her a brownie lol", 
              "so trash wouldnt do colors better tweet", 
              "love monkey brownie as much as a tweet",
              "monkey get this tweet around i think",
              "saw a brownie to make me some monkey" ])

>>> B = ' '.join(A).split()
>>> top_n_words, top_n_words_count = zip(*Counter(B).most_common(3))
>>> top_n_words_relative_count = np.array(tom_n_words_count)/len(B)
>>> top_n_words
('a', 'brownie', 'tweet')
>>> top_n_words_relative_count
array([0.07142857, 0.07142857, 0.07142857])

如果要格式化：

>>> [f"{count}/{len(B)}" for count in top_n_words_count]
['3/42', '3/42', '3/42']

或者，如果您转到pandas，则可以使用value_counts和nlargest：

>>> import pandas as pd
>>> B = pd.Series(' '.join(A).split())
>>> B = B.value_counts(normalize=True).nlargest(3)
monkey    0.071429
a         0.071429
tweet     0.071429
dtype: float64
>>> B.index.tolist()
['monkey', 'a', 'tweet']
>>> B.values.tolist()
[0.07142857142857142, 0.07142857142857142, 0.07142857142857142]

从Python中的一组句子中查找最常见单词的相对计数

1 个答案: