使用机器学习进行气象数据分析

1. 首先咱们须要找到数据，不少地方提供了api，好比：

https://www.wunderground.com/weather/api（可是这个网站不提供空气质量）

空气质量可参考：https://www.zhihu.com/question/20939327

2. 从api获取数据，使用python，代码粘出来：

# -*- coding: UTF-8 -*-
import urllib2
import json
from datetime import datetime
import pandas as pd

''''
最终选择的特征有：气温tempm, 露点dewptm, 湿度humidity, 风力wspdm, 能见度vism, 气压pressurei, 降水precipm
其中，tempm:min max mean, dewptm:min max mean, humidity:humidity, wspdm: min max, vism: mean min max, 
pressurei: max min mean, precipm:precipm
目标：fog（雾霾）
'''
def getdata(month,day,meant,meand,humi,maxw,meanv,meanp,preci,fo):
    date = datetime(2017, month, day)
    print day
    target = 'http://api.wunderground.com/api/{Your Key}/history_{}/q/CN/zmw:00000.1.54511.json?v=wuiapp'
    f = urllib2.urlopen(target.format(date.strftime('%Y%m%d')))
    json_string = f.read()
    parsed_json = json.loads(json_string)
    day = parsed_json['history']['dailysummary']
    temp = day[0]['meantempm']  #气温状况
    dewptm = day[0]['meandewptm']  # 露点状况
    hum = day[0]['humidity']  # 湿度状况
    wspdm = day[0]['maxwspdm']  # 风力状况
    vism = day[0]['meanvism']  # 能见度状况
    press = day[0]['meanpressurei']  # 气压状况
    prec = day[0]['precipm']  # 降水状况
    fog = day[0]['fog']  #雾霾状况
    meant.append(temp)
    meand.append(dewptm)
    humi.append(hum)
    maxw.append(wspdm)
    meanv.append(vism)
    meanp.append(press)
    preci.append(prec)
    fo.append(fog)
    f.close()
if __name__ == '__main__':
    meantempm = []
    meandewptm = []
    humidity = []
    maxwspdm = []
    meanvism = []
    meanpressurei = []
    precipm = []
    f = []
    for day in range(1,31):
        getdata(4, day, meantempm, meandewptm, humidity, maxwspdm, meanvism, meanpressurei, precipm, f)
        print meantempm
    #head = [u'温度',u'露点',u'湿度',u'风力',u'能见度',u'气压',u'降水',u'雾霾']
    value = [meantempm, meandewptm, humidity, maxwspdm, meanvism, meanpressurei, precipm, f]
    value = list(zip(*value))
    dataframe = pd.DataFrame(value)
    dataframe.to_csv('/Users/purixingtei/Downloads/output-2.csv', index=False, encoding="utf-8")

其中的Your Key须要被替换成本身的app key，而后主函数的循环，须要根据本身的月-日进行选择。

特别注意一点就是：不要起csv.py的名！！！

（loading）