一 先篩選出還有'from'列中帶有'iphone 6s'的行，然后對這些數據進行groupby，結果倒序排
約等同於sql中的groupby+where+order by +desc
df[df['from'].str.contains('iphone 6s plus')].groupby(['from','to'])['uid'].agg({'uv':'count'}).sort_values(by='uv',ascending=0)
篩選groupby之后排序，分組取top值（分組排序的迂回方法，不知道有沒有更好的方法）
df[df['from'].str.contains('oppo r9')].groupby(['from','to'])['uid'].agg({'uv':'count'}).sort_values(by='uv',ascending=0)['uv'].groupby(level=0,group_keys=False).nlargest(5000).to_csv('/Users/cici/Documents/group_huanji.csv',encoding='utf-8')
二 輸出A列和B列帶有某字符串的C列
df[(df['from']=='蘋果-iphone 6s') & (df['to']=='蘋果-iphone 7')]['uid'].to_csv('/Users/cici/Documents/iphone6_ip7.csv',header=0,index=False)