|
|
# -*- coding: utf-8 -*-
|
|
|
"""
|
|
|
Created on Tue Dec 28 14:59:48 2021
|
|
|
|
|
|
@author: dell
|
|
|
"""
|
|
|
|
|
|
##计算机与人工智能大作业汇报
|
|
|
|
|
|
|
|
|
# 首先我们需要导入 requests 库
|
|
|
import requests
|
|
|
|
|
|
# 请求的url
|
|
|
url = "https://www.bitpush.news/covid19/"
|
|
|
# 设置请求头信息
|
|
|
headers = {
|
|
|
"user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/95.0.4638.69 Safari/537.36"
|
|
|
}
|
|
|
|
|
|
# 使用reqeusts模快发起 GET 请求
|
|
|
response = requests.get(url, headers=headers)
|
|
|
|
|
|
# 获取请求的返回结果
|
|
|
html = response.text
|
|
|
|
|
|
# 导入 lxml
|
|
|
from lxml import etree
|
|
|
|
|
|
# 创建一个 lxml 对象，编码方式设为 utf-8
|
|
|
parse = etree.HTMLParser(encoding='utf-8') # 添加编码
|
|
|
|
|
|
# 解析 requests 返回的响应结果
|
|
|
doc = etree.HTML(html)
|
|
|
|
|
|
# 地区
|
|
|
area = doc.xpath('(//div[@class="table_container"])[2]//tbody/tr/td/span/text()')
|
|
|
|
|
|
# 确诊人数
|
|
|
person = doc.xpath('(//div[@class="table_container"])[2]//tbody/tr/td[2]/text()')
|
|
|
|
|
|
# 由于确诊人数中有逗号，我们使用列表推导式删除
|
|
|
person = [x.replace(",", "") for x in person]
|
|
|
|
|
|
# 死亡人数
|
|
|
death = doc.xpath('(//div[@class="table_container"])[2]//tbody/tr/td[3]/text()')
|
|
|
|
|
|
# 同样使用列表推导式删除逗号
|
|
|
death = [x.replace(",", "") for x in death]
|
|
|
|
|
|
# 打包数据之后将其转换成列表
|
|
|
message = list(zip(area, person, death))
|
|
|
print(message)
|
|
|
|
|
|
# 导入python中的内置模块csv
|
|
|
import csv
|
|
|
|
|
|
#用文件的方式打开
|
|
|
with open("content.csv", "w",encoding='utf-8-sig') as df:
|
|
|
w = csv.writer(df)
|
|
|
w.writerows(message)
|
|
|
#测试读取内容
|
|
|
with open("content.csv",'r',encoding='utf-8') as df:
|
|
|
data=df.read()
|
|
|
print(data)
|
|
|
|
|
|
#用pandas的方式读取数据并处理
|
|
|
import pandas as pd
|
|
|
|
|
|
# 读取数据
|
|
|
df = pd.read_csv("content.csv", names=["area", "person", "death"])
|
|
|
#给列命名
|
|
|
df.head()
|
|
|
|
|
|
df.info()
|
|
|
|
|
|
# 取前15组数据，降序排列
|
|
|
df1 = df.head(15)
|
|
|
|
|
|
import matplotlib.pyplot as plt
|
|
|
|
|
|
# 在jupyter中直接展示图像
|
|
|
#%matplotlib inline
|
|
|
# 设置中文显示
|
|
|
plt.rcParams['font.sans-serif'] = ['SimHei']
|
|
|
plt.rcParams['figure.figsize'] = (10, 5) # 设置figure_size尺寸
|
|
|
|
|
|
# x轴坐标
|
|
|
x = df1["area"].values
|
|
|
# y轴坐标
|
|
|
y = df1["person"].values
|
|
|
# 绘制柱状图
|
|
|
plt.bar(x, y,color='y')
|
|
|
plt.grid()
|
|
|
# 设置x轴名称
|
|
|
plt.xlabel("地区",fontsize=14)
|
|
|
# 设置x轴名称
|
|
|
plt.ylabel("确诊人数",fontsize=14)
|
|
|
plt.show() |
