From b8587416a3445b239c27fbf28a79413af621ebac Mon Sep 17 00:00:00 2001
From: piofyvqg4 <2825myy@163.com>
Date: Wed, 7 Dec 2022 16:46:45 +0800
Subject: [PATCH] ADD file via upload

---
 美国疫情爬取.py | 100 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 100 insertions(+)
 create mode 100644 美国疫情爬取.py

diff --git a/美国疫情爬取.py b/美国疫情爬取.py
new file mode 100644
index 0000000..5aa4abf
--- /dev/null
+++ b/美国疫情爬取.py
@@ -0,0 +1,100 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Thu Dec  1 11:48:04 2022
+
+@author: 2602103447
+"""
+
+##### 代码窗口
+#导入库
+from bs4 import BeautifulSoup
+import requests
+import pandas as pd
+import matplotlib.pyplot as plt
+
+# 请求的url
+url = 'https://www.bitpush.news/covid19/'
+# 使用reqeusts模快发起 GET 请求
+result = requests.get(url)
+#逐一解析数据
+doc = BeautifulSoup(result.text, 'html.parser')
+tbody = doc.find_all("tbody")
+us = tbody[1]
+table = us.find_all('tr')
+s = []
+c = []
+d = []
+
+for tr in table:
+    #爬取州
+    state = tr.find('span').string
+    #爬取确诊病例
+    cases = tr.find(attrs={'class':'table_card_cell_col_1 table_card_cell_int_type'}).string.replace(',','')
+    #爬取死亡人数
+    deaths = tr.find(attrs={'class':'table_card_cell_col_2 table_card_cell_float_type'}).string.replace(',','')
+    s.append(state)
+    c.append(int(cases))
+    d.append(int(deaths))
+#查找数据
+df = pd.DataFrame(list(zip(s, c, d)),
+columns =['州', '确诊病例', '死亡人数'])
+df1 = df.sort_values(by=['确诊病例'], ascending=False).head(15)
+df2 = df.sort_values(by=['死亡人数'], ascending=False).head(15)
+print('确诊病例top15排行')
+print(df1)
+print('死亡病例top15排行')
+print(df2)
+#添加搜索州的疫情情况
+state = input("请输入你要查询的州的名称： ")
+for i in df.values:
+    
+    if i[0] == state:
+        print("查询结果如下：")
+        print('州名称：',end =' ')
+        print(i[0])
+        print('确诊人数：',end =' ')
+        print(i[1])
+        print('死亡人数：',end =' ')
+        print(i[2])
+        print('查询完成')
+        break
+else:
+    print("抱歉，没有查到相关的信息。")
+
+
+#数据可视化
+%matplotlib inline
+plt.rcParams['font.sans-serif'] = ['SimHei']
+plt.rcParams['figure.figsize'] = (16,5)
+#绘制确诊人数柱形图
+x = df1['州'].values
+y = df1['确诊病例'].values
+plt.bar(x,y,width=0.6)
+plt.xlabel('州',fontsize = 14)
+plt.ylabel('确诊病例',fontsize = 14)
+plt.title('疫情确诊人数排名 top15')
+plt.show()
+#绘制死亡人数柱形图
+x = df2['州'].values
+y = df2['死亡人数'].values
+plt.bar(x,y,width=0.6)
+plt.xlabel('州',fontsize = 14)
+plt.ylabel('死亡人数',fontsize = 14)
+plt.title('死亡人数人数排名 top15')
+plt.show()
+
+#绘制确诊人数top15饼图
+labels = df1['州'].values
+oppcy = df1['确诊病例'].values
+plt.pie(oppcy,labels=labels,autopct='%1.2f%%',radius=2)
+plt.savefig("确诊人数top15饼图.jpg",dpi=200)
+plt.title('确诊人数top15饼图')
+plt.show()
+
+#绘制死亡人数top15饼图
+labels = df2['州'].values
+oppcy = df2['死亡人数'].values
+plt.pie(oppcy,labels=labels,autopct='%1.2f%%',radius=2)
+plt.savefig("死亡人数top15饼图.jpg",dpi=200)
+plt.title('死亡人数top15饼图')
+plt.show()
\ No newline at end of file