一、概述

1.数据来源

2.数据概述

3.依赖包及格式文件介绍

二、弹幕数据分析及可视化

1.导入全部依赖包

import requests
import re
import json
import jieba
from wordcloud import WordCloud
import matplotlib.pyplot as plt
import pandas as pd

2.设置matplotlib中文字体

plt.rcParams['font.sans-serif'] = ['simhei']  # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号

3.通过requests来获取弹幕内容数据

url = '<https://mfm.video.qq.com/danmu?otype=json&callback=jQuery19109073533539243266_1625488331995&target_id=6972946900%26vid%3Dq00389hmv7w&session_key=0%2C0%2C0&timestamp=15&_=1625488331996>'
response = requests.get(url)
response.encoding = 'utf-8'

4.从包含回调函数的字符串中提取JSON格式内容数据

main_json = re.findall('jQuery\\d+_\\d+[(](.*)[)]',response.text)

5.加载JSON数据解析并进行数据整理

js = json.loads(main_json[0])
danmu = [''.join(item['content'].split()) for item in js['comments']]
data = [[item['timepoint'],''.join(item['content'].split())] for item in js['comments']]

6.保存数据到CSV文件中

def save_csv(data):
    for item in data:
        with open('danmu.csv','a',encoding='utf-8') as f:
            f.write(','.join([str(i) for i in item])+'\\n')

7.统计弹幕词频并绘图

words = jieba.cut(' '.join(danmu),cut_all=True)
words = movestopwords(words)
word_hot={}
#统计词频
for i in words:
    if len(str(i))>1:
        word_hot[i] = word_hot.get(i,0)+1
# 词频按从大到小进行排序
word_hot = sorted(word_hot.items(),  key=lambda d: d[1], reverse=True)[:20]
word_hot = [[i[0] for i in word_hot],[i[1] for i in word_hot]]
plt.figure(figsize=(15,5))
plt.bar(word_hot[0],word_hot[1])
plt.show()

8.绘制词云图

wc = WordCloud(
    font_path='simhei.ttf',     #字体路劲
    background_color='white',   #背景颜色
    width=1000,
    height=600,
    max_font_size=300,            #字体大小
    min_font_size=10,
    max_words=100,
    collocations=False,
    font_step=1
)
#这里云图 传入数据必须是带有空格的才能识别
s=wc.generate(' '.join(words))
plt.figure(figsize=(10,6))
#必须有这个才会显示图
plt.imshow(s)
#去除横纵坐标轴
plt.axis('off')
plt.show()

9.统计弹幕中出现名字的数量并绘图

names = [ 'baby','周笔畅','蔡徐坤','李沁','宋雨琦','谭松韵','松韵','肉粽鹅','黄旭熙']
for name in names:
    jieba.add_word(name,freq=1)
words = jieba.cut(' '.join(danmu))
words = movestopwords(words)
word_hot = {}
for i in names:
    word_hot[i] = words.count(i)
word_hot = sorted(word_hot.items(),  key=lambda d: d[1], reverse=True)[:20]
word_hot = [[i[0] for i in word_hot],[i[1] for i in word_hot]]
plt.figure(figsize=(15,5))
plt.bar(word_hot[0],word_hot[1])
plt.show()
wc = WordCloud(
    font_path='simhei.ttf',     #字体路劲
    background_color='white',   #背景颜色
    width=1000,
    height=600,
    max_font_size=300,            #字体大小
    min_font_size=10,
    max_words=100,
    collocations=False,
    font_step=1
)
s=wc.generate(' '.join([item for item in words if item in names]))
plt.figure(figsize=(10,6))
plt.imshow(s)
plt.axis('off')
plt.show()

10.统计时间上的弹幕分布

df = pd.read_csv('danmu.csv',names=['time','content'])
time = range(30)
count_df = df['time'].value_counts(sort=True)
count_dic = dict(count_df.items())
count = []
for item in time:
    count.append(count_dic[item])
plt.bar(time,count)
plt.show()

三、结论