python - groupby后如何设置聚合？

Question

鉴于我有一个数据集如下:

dt = {
    "facility":["Ann Arbor","Ann Arbor","Detriot","Detriot","Detriot"],
    "patient_ID":[4388,4388,9086,9086,9086],
    "year":[2004,2007,2007,2008,2011],
    "month":[8,9,9,6,2],
    "Nr_Small":[0,0,5,12,10],
    "Nr_Medium":[3,1,1,4,3],
    "Nr_Large":[2,0,0,0,0],
    "PeriodBetween2Visits" : [10,0,12,3,1],
    "NumberOfVisits" : [2,2,3,3,3]

}

dt = pd.DataFrame(dt)

我需要保留 groupby patient_ID ，然后保留 facility , patient_ID , NumberOfVisits ，但是 最大值 和 最低 的 PeriodBetween2Visits .

这是我尝试过的:

dt = dt.groupby(['patient_ID'],as_index=False)["facility","patient_ID","PeriodBetween2Visits","NumberOfVisits"].agg({'PeriodBetween2Visits': ['min', 'max']})


dt.head()

但是，这不是我需要的!

对我来说正确的输出如下:

Answer 1

我在这里使用命名聚合，它内置于 groupby 和 agg recently :

 dt.groupby(['facility','patient_ID']).agg(
                 Min_PeriodBetween2Visits=('PeriodBetween2Visits','min'),
                 Max_PeriodBetween2Visits=('PeriodBetween2Visits','max'),
                 NumberOfVisits=('NumberOfVisits','nunique')).reset_index()

    facility  patient_ID  Min_PeriodBetween2Visits  Max_PeriodBetween2Visits  \
0  Ann Arbor        4388                         0                        10   
1    Detriot        9086                         1                        12   

   NumberOfVisits  
0               2  
1               3