day1-review

실습 코드

import pandas as pd
df = pd.read_excel('patents.xlsx')

!pip install kiwipiepy

from kiwipiepy import Kiwi
kiwi = Kiwi()

def extract_nouns(text):
    result = kiwi.tokenize(text)
    for token in result:
        if token.tag in ['NNG', 'NNP']:
            yield token.form

from sklearn.feature_extraction.text import CountVectorizer
cv = CountVectorizer(
    max_features=100,       # 최대 단어 수(빈도 순)
    tokenizer=extract_nouns) # 토큰화 방법

dtm = cv.fit_transform(df['abstract'])

word_count = pd.DataFrame({
    '단어': cv.get_feature_names_out(),
    '빈도': dtm.sum(axis=0).flat
})
count_dic = dict(zip(word_count.단어, word_count.빈도))

word_count.sort_values('빈도', ascending=False) # 정답: 발명

!pip install wordcloud

from wordcloud import WordCloud
from PIL import Image
import numpy as np
from wordcloud import ImageColorGenerator

# 모양 잡기
mask = Image.open('mask.png') # .convert('L') 없음
mask = np.asarray(mask)
wc = WordCloud(font_path='온글잎 누카.ttf', background_color='white', mask=mask)
wc.fit_words(count_dic)
 
# 색칠하기
color_func = ImageColorGenerator(mask)
cloud = wc.recolor(color_func=color_func)
cloud.to_image()