pandas排序

#分组逐月求和dataNew = data2.groupby(['shop_id'])[['sale_amt']].resample('M',how = sum) #分组逐日求最值dataMax = data2.groupby(['shop_id'])[['sale_amt']].resample('D',how = max)

#以shop_id分组，返回sale_amt最大的对应索引day_Max = data2.groupby(['shop_id'])[['sale_amt']].idxmax()

#按照不同shop_id分别排序再拼接成同一张表start = data2.loc[data2['shop_id'] == 1].sort_values(['sale_amt'],ascending = False)for i in range(2,3001):    mindi = data2.loc[data2['shop_id'] == i].sort_values(['sale_amt'],ascending = False)    start = pd.concat([start,mindi],axis=0)

最后只想出这种方法，肯定还有更好的办法实现。