Python：如何使用自定义顺序对分组数据集重新编制索引

我已经按照我想要的方式对数据进行了分组，但是月份不符合顺序

sign_off=df1.groupby(['Sign off','LOB']).sum()
print(sign_off)

导致

                          Test Cases
Sign off     LOB                    
April2019    Sales            135
April2020    Systems          36
             Others           49
August2019   Systems          13
             Sales            414
             DevOps           47
February2019 Systems          42
February2020 Systems          76
             Sales            151
January2019  ECS              251
             Systems          157
             Sales            116
July2019     Systems          45
             Sales            9
June2019     Systems          164
March2019    ECS              37
             Systems          181
March2020    Systems          13
May2019      Systems          7
May2020      Systems          249
             Others           60
November2019 Systems          49
October2019  Systems          479
             Sales            130

这就是我希望它的格式，但月份是按字母顺序排列的（我希望它们保持这种字符串格式）。因此，现在我需要重新安排签准月份，我尝试了以下方法：

order = ['January2019','February2019','March2019','April2019','May2019','June2019','July2019','August2019','October2019','November2019','February2020','March2020','April2020','May2020']
sign_off.reindex(order)

这将导致错误：

TypeError:Expected tuple，got str

---

我需要它根据月份列对数据集进行重新排序。我希望月份的顺序符合我指定的顺序，并具有与该月份相关的正确业务线和测试用例。

您可以尝试此操作，在groupby上使用

as_index=False

将列从索引中删除，然后执行其余操作：

sign_off=df1.groupby(['Sign off','LOB'], as_index=False).sum()
df['Sign off'] = pd.to_datetime(df['Sign off'], format='%B%Y', errors='coerce').dt.strftime('%m%Y')
df.sort_values(by=['Sign off'], inplace=True)
df['Sign off'] = pd.to_datetime(df['Sign off']).dt.strftime('%B%Y')
print(df)

输出：

       Sign off      LOB  Test Cases
9   January2019      ECS       251.0
15    March2019      ECS        37.0
17    March2020  Systems        13.0
0     April2019    Sales       135.0
1     April2020  Systems        36.0
18      May2019  Systems         7.0
19      May2020  Systems       249.0
14     June2019  Systems       164.0
12     July2019  Systems        45.0
3    August2019  Systems        13.0
22  October2019  Systems       479.0

---

您可以尝试这样做，在groupby上使用

as_index=False

从索引中取出列，然后执行其余操作：

sign_off=df1.groupby(['Sign off','LOB'], as_index=False).sum()
df['Sign off'] = pd.to_datetime(df['Sign off'], format='%B%Y', errors='coerce').dt.strftime('%m%Y')
df.sort_values(by=['Sign off'], inplace=True)
df['Sign off'] = pd.to_datetime(df['Sign off']).dt.strftime('%B%Y')
print(df)

输出：

       Sign off      LOB  Test Cases
9   January2019      ECS       251.0
15    March2019      ECS        37.0
17    March2020  Systems        13.0
0     April2019    Sales       135.0
1     April2020  Systems        36.0
18      May2019  Systems         7.0
19      May2020  Systems       249.0
14     June2019  Systems       164.0
12     July2019  Systems        45.0
3    August2019  Systems        13.0
22  October2019  Systems       479.0