如何将权重（频率）不同的词语通过词语大小的形式可视化较为重要。这里使用了WordCloud进行展示。

1. 词语频率作为“词语大小”展示

如果给定的是一段text，从里面分词+统计频率展示词语大小，那么使用到的是generate。
dzdpstopwords.txt
d_text1.csv

from wordcloud import WordCloud
import matplotlib.pyplot as plt
import re
import pandas as pd
import numpy as np
#导入停止词
def get_stop_word(url):
    stop_word = []
    with open(url, encoding='UTF-8') as fp:
        for i in fp:
            zh = re.compile("[^\u4e00-\u9fa5]") #匹配不是中文的字符
            i = zh.sub('', i)
            stop_word.append(i)        
    return stop_word
def segmentWord(cont, stop_word):
    c = [] 
    for i in cont:
        zh = re.compile("[^\u4e00-\u9fa5]") #匹配不是中文的字符
        string1 = zh.sub('', i)             #将string1中匹配到的字符替换成空字符
        a = list(jieba.cut(string1))
        #删去停止词
        a1 = [item for item in a if item not in stop_word]
        b = ' '.join(a1)
        c.append(b)
    return c
# 导入停止词+需要分析的文本，分完词+剔除停止词，再用空格连成一个string。
d_train=pd.read_csv(r'/trainingset.csv',engine='python', encoding="utf_8")# 训练数据集
stop_word_dp = get_stop_word(r'/dzdpstopwords.txt')
text = segmentWord(d_train['content'], stop_word_dp)
text1 = " ".join(text)
# 通过generate对string统计词频+生成大小；
# font_path是显示指定字体，
# width和height是图片的长款，
# background_color是背景色，这里我选择了白色。
# color_func是制定字体显示的同一颜色，通过lambda指定了RGB。
# relative_scaling=1，将字体大小严格按照频率展示。
wc1 = WordCloud(font_path=r'\STZHONGS.TTF', 
                      width=4000,
                      height=8000,             
                      background_color='#FFFFFF',
                      color_func=lambda *args, **kwargs: (210,10,16), 
                      relative_scaling=1                      
                      ).generate(text1)
plt.rcParams['savefig.dpi'] = 1000 #图片像素
plt.rcParams['figure.dpi'] = 1000 #分辨率
plt.imshow(wc1, cmap='prism', interpolation='gaussian')
plt.axis("off")

2.词语权重作为“词语大小”展示

如果仅仅给定了每个词语的权重，那么使用的是generate_from_frequancy。

word_E_dict = {'we':0.9, 'are':0.1, 'good': 0.2}
wordcloud = WordCloud(font_path=r'\STZHONGS.TTF', 
                      width=6000,
                      height=8000,             
                      background_color='#FFFFFF',
                      color_func=lambda *args, **kwargs: (227,82,4), 
                      relative_scaling=1                      
                      )
wordcloud.generate_from_frequencies(word_E_dict)
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.show()

3. 如果要字符都竖着

那么直接修改原始的字符，将每个string中的每个character后面加上’\n’。
word.xlsx

import pandas as pd
import numpy as np
from wordcloud import WordCloud
import matplotlib.pyplot as plt
word = pd.read_excel(r'\word.xlsx')
word_E = word[word['类别'] == '环境（E）'] 
word_S = word[word['类别'] == '社会（S）'] 
word_G = word[word['类别'] == '治理（G）'] 
word_E_dict = dict()
word_S_dict = dict()
word_G_dict = dict()
for i, j in zip(word_E['指标名称'], word_E['2020年披露率']):
    iii = ''
    for ii in i:
        iii += ii + '\n'
    iii = iii[:-1]
    word_E_dict[iii] = j
for i, j in zip(word_S['指标名称'], word_S['2020年披露率']):
    iii = ''
    for ii in i:
        iii += ii + '\n'
    iii = iii[:-1]
    word_S_dict[iii] = j
for i, j in zip(word_G['指标名称'], word_G['2020年披露率']):
    iii = ''
    for ii in i:
        iii += ii + '\n'
    iii = iii[:-1]
    word_G_dict[iii] = j
plt.rcParams['savefig.dpi'] = 4000 #图片像素
plt.rcParams['figure.dpi'] = 4000 #分辨率
wordcloud = WordCloud(font_path=r'STZHONGS.TTF', 
                      width=6000,
                      height=8000,             
                      background_color='#FFFFFF',
                      color_func=lambda *args, **kwargs: (227,82,4), 
                      relative_scaling=1                      
                      )
wordcloud.generate_from_frequencies(word_E_dict)
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.show()