1. 数据采集逻辑

  1. 数据schema

历史天气数据schema

{

‘当日信息’:’2023-01-01 星期日’,

‘最高气温’: 8℃”,

‘最低气温’: ‘5℃’,

‘天气’: ‘多云’,

‘风向信息’:’北风 3级’

}

  1. 数据爬取

1.导入库

import numpy as npimport pandas as pdimport requestsfrom bs4 import BeautifulSoupfrom matplotlib import pyplot as pltfrom pandas import Series, DataFrame

2.对程序进行伪装

headers = {'Host': 'lishi.tianqi.com','user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36 Edg/110.0.1587.63'}

3.抓取天气数据

url = 'https://lishi.tianqi.com/shanghai/202301.html'# 上海 2023年1月天气res = requests.get(url, headers=headers)res.encodind = 'utf-8'html = BeautifulSoup(res.text, 'html.parser')data_all = []tian_three = html.find("div", {"class": "tian_three"})lishi = tian_three.find_all("li")for i in lishi:lishi_div = i.find_all("div")data = []for j in lishi_div:data.append(j.text)data_all.append(data)print(data_all)

4.数据存储

在数据存储前,对数据进行处理,便于后期的数据分析。将上面的“当天信息”字段拆分为“日期”和“星期”两个字段,“风向信息”也是如此。最后,将数据保存为csv文件中。

weather = pd.DataFrame(data_all)weather.columns = ["当日信息", "最高气温", "最低气温", "天气", "风向信息"]weather_shape = weather.shapeprint(weather)weather['当日信息'].apply(str)result = DataFrame(weather['当日信息'].apply(lambda x: Series(str(x).split(' '))))result = result.loc[:, 0:1]result.columns = ['日期', '星期']weather['风向信息'].apply(str)result1 = DataFrame(weather['风向信息'].apply(lambda x: Series(str(x).split(' '))))result1 = result1.loc[:, 0:1]result1.columns = ['风向', '级数']weather = weather.drop(columns='当日信息')weather = weather.drop(columns='风向信息')weather.insert(loc=0, column='日期', value=result['日期'])weather.insert(loc=1, column='星期', value=result['星期'])weather.insert(loc=5, column='风向', value=result1['风向'])weather.insert(loc=6, column='级数', value=result1['级数'])weather.to_csv("上海23年1月天气.csv", encoding="utf_8")

5.数据分析

注:数据分析用的是北京2023年1月的天气数据,如下图:

1.2023北京1月天气情况

# 数据处理plt.rcParams['font.sans-serif'] = ['SimHei']plt.rcParams['axes.unicode_minus'] = Falseweather['最高气温'] = weather['最高气温'].map(lambda x: int(x.replace('℃', '')))weather['最低气温'] = weather['最低气温'].map(lambda x: int(x.replace('℃', '')))dates = weather['日期']highs = weather['最高气温']lows = weather['最低气温']# 画图fig = plt.figure(dpi=128, figsize=(10, 6))plt.plot(dates, highs, c='red', alpha=0.5)plt.plot(dates, lows, c='blue', alpha=0.5)plt.fill_between(dates, highs, lows, facecolor='blue', alpha=0.2)# 图表格式# 设置图标的图形格式plt.title('2023北京1月天气情况', fontsize=24)plt.xlabel('', fontsize=6)fig.autofmt_xdate()plt.ylabel('气温', fontsize=12)plt.tick_params(axis='both', which='major', labelsize=10)# 修改刻度plt.xticks(dates[::5])# 显示plt.show()

2.北京23年1月天气候分布饼图

2023年一月份有31天,循环遍历时注意循环次数。

# 天气可视化饼图weather = list(weather['天气'])dic_wea = {}for i in range(0, 31):if weather[i] in dic_wea.keys():dic_wea[weather[i]] += 1else:dic_wea[weather[i]] = 1print(dic_wea)explode = [0.01] * len(dic_wea.keys())color = ['lightskyblue', 'silver', 'yellow', 'salmon', 'grey', 'lime', 'gold', 'red', 'green', 'pink']plt.pie(dic_wea.values(), explode=explode, labels=dic_wea.keys(), autopct='%1.1f%%', colors=color)plt.title('北京23年1月天气候分布饼图')plt.show()

3.风级图

自定义change_wind函数,将风向信息转换为数值,并计算出各风向的风速平均值。

def change_wind(wind):"""改变风向"""for i in range(0, 31):if wind[i] == "北风":wind[i] = 90elif wind[i] == "南风":wind[i] = 270elif wind[i] == "西风":wind[i] = 180elif wind[i] == "东风":wind[i] = 360elif wind[i] == "东北风":wind[i] = 45elif wind[i] == "西北风":wind[i] = 135elif wind[i] == "西南风":wind[i] = 225elif wind[i] == "东南风":wind[i] = 315return wind# 风向雷达图wind = list(weather['风向'])weather['级数'] = weather['级数'].map(lambda x: int(x.replace('级', '')))# weather['级数']=pd.to_numeric(weather['级数'])wind_speed = list(weather['级数'])wind = change_wind(wind)degs = np.arange(45, 361, 45)temp = []for deg in degs:speed = []# 获取 wind_deg 在指定范围的风速平均值数据for i in range(0, 31):if wind[i] == deg:speed.append(wind_speed[i])if len(speed) == 0:temp.append(0)else:temp.append(sum(speed) / len(speed))print(temp)N = 8theta = np.arange(0. + np.pi / 8, 2 * np.pi + np.pi / 8, 2 * np.pi / 8)# 数据极径radii = np.array(temp)# 绘制极区图坐标系plt.axes(polar=True)# 定义每个扇区的RGB值(R,G,B),x越大,对应的颜色越接近蓝色colors = [(1 - x / max(temp), 1 - x / max(temp), 0.6) for x in radii]plt.bar(theta, radii, width=(2 * np.pi / N), bottom=0.0, color=colors)plt.title('风级图', x=0.2, fontsize=20)plt.show()