前言:
当前咱们对“基于python的影评数据分析”可能比较着重,朋友们都想要剖析一些“基于python的影评数据分析”的相关资讯。那么小编同时在网络上搜集了一些有关“基于python的影评数据分析””的相关资讯,希望你们能喜欢,各位老铁们快快来学习一下吧!本次通过猫眼电影,对春节贺岁大片【满江红】进行数据分析。而本次我们通过动态接口形式获取评论信息,静态HTML解析需要额外的字体解析,网上的教程也已经很全了,有兴趣的小伙伴们也可以多多冲浪或和本人探讨哈!
一、 接口分析
1. 目标站点:猫眼H5
2. 通过滑动查看评论信息,或点击评论进入评论子页面滑动,即可抓取到相关接口(浏览器F12工具中只能抓取到子评论接口,如果要整个评论的需要抓包工具配合或使用手机抓包)
3. 评论接口(已加密处理)
aHR0cHM6Ly9tLm1hb3lhbi5jb20vYXBvbGxvL2Fwb2xsb2FwaS9tbWRiL3JlcGxpZXMvY29tbWVudC8xMTY3MTI5MDg5Lmpzb24/X3ZfPXllcyZvZmZzZXQ9NDA=
二、 响应分析通过子评论接口,可以分析出来相关字段(昵称、性别、评分、评论内容、评论点赞量、用户等级等)
{ "cmts": [ { "approve": 0, "assistAwardInfo": { "avatar": "", "celebrityId": 0, "celebrityName": "", "rank": 0, "title": "" }, "avatarurl": ";, "channelId": 70001, "content": "在电影院看真的很有氛围!背景音乐也很加分", "deleted": false, "id": 1171602285, "ipLocName": "福建", "nickName": "腿小菇", "time": "2023-02-27 10:24", "userId": 1322748722, "userLevel": 3, "vipInfo": "", "vipType": 0 } ], "ocm": { "approve": 8657, "approved": false, "assistAwardInfo": { "avatar": "", "celebrityId": 0, "celebrityName": "", "rank": 0, "title": "" }, "authInfo": "", "avatarurl": ";, "content": "刚看完满江红,真的好看,这是我看过最值的一部电影,反转反转再反转,真的是永远想不到下一步是什么,而且还很搞笑,搞笑又宏伟,真的描述不出来这个电影的好,都给我去看!满江红!入股不亏!!!!", "id": 1167129089, "ipLocName": "辽宁", "isMajor": false, "juryLevel": 0, "majorType": 0, "mvid": 1462626, "nick": "Gpc126688235", "nickName": "Gpc126688235", "oppose": 0, "pro": false, "reply": 680, "score": 5, "spoiler": 0, "supportComment": true, "supportLike": true, "sureViewed": 1, "tagList": { "fixed": [ { "id": 1, "name": "购票好评" }, { "id": 4, "name": "购票" }, { "id": 6, "name": "优质评价" } ] }, "time": "2023-01-22 12:19", "userId": 3164097169, "userLevel": 2, "videoDuration": 0, "vipInfo": "", "vipType": 0 }, "total": 60}
2. 完整comment接口响应示例
{ "data": { "hotIds": [ 1167280609, 1167187803 ], "total": 16521, "comments": [ { "avatarUrl": ";, "buyTicket": false, "channelId": 3, "content": "还行吧,没有看开心 ", "delete": false, "follow": false, "gender": 1, "id": 1171756165, "imageUrls": [], "ipLocName": "山东", "likedByCurrentUser": false, "major": false, "movie": { "id": 0, "sc": 0 }, "movieId": 1462626, "nick": "淘嘉豪", "replyCount": 0, "score": 9, "showApprove": false, "showVote": false, "spoiler": false, "startTime": "1677923460000", "tagList": [ { "id": 1, "name": "购票好评" }, { "id": 4, "name": "购票" } ], "time": 1677923460000, "ugcType": 11, "upCount": 0, "userId": 71317227, "userLevel": 2, "vipType": 0 }, ], "t2total": 0, "myComment": {} }, "paging": {}, "ts": 1677956823197}三、数据解析构造请求头,模拟数据请求
def get_film_data(offset = 0, filename="film"): url = f'aHR0cHM6Ly9tLm1hb3lhbi5jb20vYXBvbGxvL2Fwb2xsb2FwaS9tbWRiL3JlcGxpZXMvY29tbWVudC8xMTY3MTI5MDg5Lmpzb24/X3ZfPXllcyZvZmZzZXQ9NDA=' headers = { 'User-Agent': 'Mozilla/5.0 (iPhone; CPU iPhone OS 11_0 like Mac OS X) AppleWebKit/604.1.38 (KHTML, like Gecko) Version/11.0 Mobile/15A372 Safari/604.1' } cookies = { 'uuid_n_v':'v1', 'iuuid':'942C12B0DF4311E9ADA9C1C3B540BA45F066B2B3028841B8A0BC3544E4C0AD17', 'ci':'1%2C%E5%8C%97%E4%BA%AC', '_lxsdk_cuid':'16d6c9b401ec8-0c6c86354bd8a9-5b123211-100200-16d6c9b401ec8', 'webp':'true', '_lxsdk':'942C12B0DF4311E9ADA9C1C3B540BA45F066B2B3028841B8A0BC3544E4C0AD17' } # 开始页面请求,返回响应内容 response = requests.get(url,headers=headers,cookies=cookies).json() # 总评论数 total = response['total'] print(total) # 评论信息列表 cmts = response['cmts'] pprint(cmts) for comment in cmts: data = [] # 评论id # id = comment['id'] # 评论内容 content = comment['content'] # 用户昵称 nickName = comment['nickName'] # 用户评分 score = comment['score'] # 评论时间 # startTime = comment['time'] # 用户id userId = comment['userId'] # 用户等级 userLevel = comment['userLevel'] # 用户性别 gender = comment.get('gender',None) data['nickName '] = nickName data['gender'] = gender data['score'] = score data['content'] = content data['userId '] = userId data['userLevel'] = userLevel save_data_csv(data,filename) return total
2. 数据存储(这里为以csv演示)
def save_data_csv(data, file_name): with open(file_name,'a',encoding='utf-8-sig',newline='')as fp: # 创建写对象 writer = csv.writer(fp) title = ['nickName ','gender','score','content','userId ','userLevel'] # 解决循环存储,表头重复问题 with open(file_name,'r',encoding='utf-8-sig',newline='')as fp: # 创建读对象 reader = csv.reader(fp) if not [row for row in reader]: writer.writerow(title) writer.writerow([data[i] for i in title]) else: writer.writerow([data[i] for i in title]) print('*'*10+'保存完毕'+'*'*10)
四、数据可视化影评分词
def wordcloud_analysis(file_name): df = pd.read_csv(file_name, encoding='utf-8') content = df['content'].to_string() # 开始分词 使用jieba进行精确分词获取词语列表 words = jieba.lcut(content) # 使用空格拼接获得字符串 words = ' '.join(words) # 生成词云 # 读取图片,生成图片形状 mask_pic = np.array(Image.open('1.jpg')) words_cloud = WordCloud( background_color='white', # 词云图片的背景颜色 width=800, height=600, # 词云图片的宽度,默认400像素;词云图片的高度,默认200像素 font_path='msyh.ttf', # 词云指定字体文件的完整路径 max_words=200, # 词云图中最大词数,默认200 max_font_size=80, # 词云图中最大的字体字号,默认None,根据高度自动调节 min_font_size# 词云图中最小的字体字号,默认4号 font_step=1, # 词云图中字号步进间隔,默认1 random_state=30, # 设置有多少种随机生成状态,即有多少种配色方案 mask=mask_pic # 词云形状,默认None,即方形图 ).generate(words) # 有jieba分词拼接的字符串生成词云 words_cloud.to_file('comment.png') # 保存词云为图片 # 使用plt显示词云 plt.imshow(words_cloud, interpolation='bilinear') # 消除坐标轴 plt.axis('off') plt.show()
2. 观看人群性别及评分占比分析(由于取得部分数据,不代表最终现实结果,勿纠)
def gender_pie_analysis(file_name): df = pd.read_csv(file_name, encoding='utf-8') print(df) # # # 1.观看人群性别 gender = df['gender'].value_counts() print(gender) # 饼图,标题:观看人群性别占比 # 调用自定义饼图函数 # 创建画布和轴 fig, ax = plt.subplots(figsize=(6, 6), dpi=100) # plt.figure() size = 0.5 # labels = data.index ax.pie(gender, labels=['女','男','未知'], startangle=90, autopct='%.1f%%' , colors=sns.color_palette('husl', len(gender)), radius=1, # 饼图半径,默认为1 pctdistance=0.75, # 控制百分比显示位置 wedgeprops=dict(width=size, edgecolor='w'), # 控制甜甜圈的宽度 textprops=dict(fontsize=10) # 控制字号及颜色 ) ax.set_title("【满江红】观看人群性别占比", fontsize=15) # plt.title(title) plt.show()
3. 用户等级分析
def user_level_bar_analysis(file_name): df = pd.read_csv(file_name, encoding='utf-8') print(df) userLevel = df['userLevel'].value_counts().sort_index() print(userLevel) x = userLevel.index y = userLevel fig, ax = plt.subplots() plt.bar(x, y, color='#DE85B5') # 柱状图标题 plt.title('评论用户等级数量分布柱状图') plt.grid(True, axis='y', alpha=1) for i, j in zip(x, y): plt.text(i, j, '%d' % j, horizontalalignment='center', ) ax.spines['right'].set_visible(False) ax.spines['top'].set_visible(False) plt.show()
该篇文章只是从评分角度去做的数据分析,其实还可以从影视类型、年度电影Top、票房等角度进一步做数据分析。
该篇文章来自本人知乎号:梓羽Python
文章链接:
标签: #基于python的影评数据分析