白天映射时不能一边旋转一边排序

Question

我正试着给时间贴上这样的标签

df['day-hour'] = ('Day' + (df['hour'] // 24).add(1).astype(str) +
              ' - ' + (df['hour'] % 24).astype(str))

所以，结果是

customer_id  hour   day-hour
1              10   Day1 - 10
1             123   Day6 - 3
1             489   Day21 - 9
2             230   Day9 - 14

然后我尝试对df.groupby(['customer_id','day-hour']).size().unstack(fill_value=0)进行分组

结果是

day-hour               Day1 - 10   Day6 - 3   Day21 - 9   Day9 - 14
customer_id
1                              1          1           1           0
2                              0          0           0           1

我所期望的输出按实际天数排序，如下所示

day-hour               Day1 - 10   Day6 - 3   Day9 - 14   Day21 - 9
customer_id
1                              1          1           0           1
2                              0          0           1           0

我该改什么代码？

Answer 1

有两种可能的解决方案--在注释中添加指向@Zero的零：

df['day-hour'] = ('Day' + (df['hour'] // 24).add(1).astype(str).str.zfill(2) +
              ' - ' + (df['hour'] % 24).astype(str).str.zfill(2) )

或按custom function with 2 fields排序

df = df[sorted(df.columns,key=lambda x: (int(x.split(' - ')[0][3:]), int(x.split(' - ')[1])))]

更具可读性：

def f(x):
    a = x.split(' - ')
    return (int(a[0][3:]), int(a[1]))

df = df[sorted(df.columns, key=f)]
print (df)
   Day1 - 10  Day6 - 3  Day9 - 14  Day21 - 9
1          1         1          0          1
2          0         0          1          0