Python 如何在日期列上创建数据透视表并计算时间差？

我有以下数据帧

D_DATE       BIN Number   Disposition    Unit Assigned        
2018-01-04    10005      SWO Issued      PLUMBING DIVISION     
2016-06-23    10005      SWO Issued      SCAFFOLD UNIT         
2016-06-23    10005      SWO Rescinded   SCAFFOLD UNIT         
2018-01-17    10005      SWO Rescinded   PLUMBING DIVISION  
2019-01-04    10006      SWO Rescinded   BEST SQUAD 
2018-12-21    10006      SWO Issued      BEST SQUAD 
2020-02-10    10006      SWO Issued      BEST SQUAD
2020-02-25    10006      SWO Rescinded   BEST SQUAD

df = pd.DataFrame({'D_DATE':['2018-01-04','2016-06-23','2016-06-23','2018-01-17','2019-01-04','2018-12-21','2020-02-10','2020-02-25'],
                    'BIN Number': ['10005', '10005', '10005', '10005', '10006','10006','10006','10006] ,
                   'Disposition': ['SWO Issued', 'SWO Issued', 'SWO Rescinded', 'SWO Rescinded','SWO Rescinded','SWO Issued','SWO Issued','SWO Rescinded'] ,
                   'Unit Assigned': ['PLUMBING DIVISION', 'SCAFFOLD UNIT', 'SCAFFOLD UNIT', 'PLUMBING DIVISION','BEST SQUAD','BEST SQUAD','BEST SQUAD','BEST SQUAD']})

如果可能的话，我想创建一个数据透视表，这样我就有两列用于日期，一列用于发布数据，另一列用于撤销日期，但在数据透视中，我需要维护单位，因此我应该有三列：

单位、发行日期、撤销日期

接下来我想计算发行日期和撤销日期之间的时间差

输出：

Unit Assigned      SWO Issued     SWO Rescinded    Time Difference
PLUMBING DIVISION  2018-01-04     2018-01-17        13 days
SCAFFOLD UNIT      2016-06-23     2016-06-23        0 days
BEST SQUAD         2018-12-21     2019-01-04        14 days
BEST SQUAD         2020-02-10     2020-02-25        15 days

Disposition     BIN Number  Unit Assigned       SWO Issued  SWO Rescinded   Time_Different
0               10005       PLUMBING DIVISION   2018-01-04  2018-01-17      13 days 
1               10005       SCAFFOLD UNIT       2016-06-23  2016-06-23      0 days
2               10006       BEST SQUAD          2018-12-21  2019-01-04      14 days
3               10006       BEST SQUAD          2020-02-10  2020-02-25      15 days

---

谢谢你的帮助。谢谢

我相信这是

pivot/pivot\u表

：

# convert to datetime if not already is
df['D_DATE'] = pd.to_datetime(df['D_DATE'])

(df.assign(idx=df.groupby(['BIN Number', 'Disposition','Unit Assigned']).cumcount())
   .pivot_table(index=['idx','BIN Number', 'Unit Assigned'], 
                columns='Disposition', 
                values='D_DATE',
                aggfunc='first')
   .reset_index()
   .assign(Time_Different=lambda x: x['SWO Rescinded'] - x['SWO Issued'])
   .drop('idx',axis=1)

)

输出：

Unit Assigned      SWO Issued     SWO Rescinded    Time Difference
PLUMBING DIVISION  2018-01-04     2018-01-17        13 days
SCAFFOLD UNIT      2016-06-23     2016-06-23        0 days
BEST SQUAD         2018-12-21     2019-01-04        14 days
BEST SQUAD         2020-02-10     2020-02-25        15 days

Disposition     BIN Number  Unit Assigned       SWO Issued  SWO Rescinded   Time_Different
0               10005       PLUMBING DIVISION   2018-01-04  2018-01-17      13 days 
1               10005       SCAFFOLD UNIT       2016-06-23  2016-06-23      0 days
2               10006       BEST SQUAD          2018-12-21  2019-01-04      14 days
3               10006       BEST SQUAD          2020-02-10  2020-02-25      15 days

---

你好谢谢，它能工作，但有一个小问题。例如，如果我有相同的单位问题&在不同的时间撤销相同仓位号的SWO，它无法计算。如果管道部门对BIN 10005发布了另一份SWO，则该SWO不会出现。它只显示第一个。帖子现在更新了。请看一看。谢谢，搞定了！！伟大的谢谢你的帮助。谢谢。