Python 使用pandas转换csv

我有一个csv文件，需要根据这两列进行转换

示例代码段：

a,UUID_ID,c,d
1,UUID1,3,2
1,UUID2,'add',2
3,UUID3,23-2-2018,5

a,UUID_ID,c,d
1,UUID1,3,  # here I put nan value for d column. So 
   # in output it should still consider that
1,UUID2,'add',   # here I put nan value for d column. So
   # in output it should still consider that
3,UUID3,23-2-2018,5

预期输出如下所示：

a,d,UUID1, UUID2, UUID3
1,2,2,'add',
3,5,,,23-2-2018

a,d,UUID1, UUID2, UUID3
1,,2,'add',
3,5,,,23-2-2018

我尝试将文件读入pandas并创建另一个dataframe副本，但不知道如何将它们连接在一起。有没有什么好办法来解决这个问题

---

跟进

示例代码段：

a,UUID_ID,c,d
1,UUID1,3,2
1,UUID2,'add',2
3,UUID3,23-2-2018,5

a,UUID_ID,c,d
1,UUID1,3,  # here I put nan value for d column. So 
   # in output it should still consider that
1,UUID2,'add',   # here I put nan value for d column. So
   # in output it should still consider that
3,UUID3,23-2-2018,5

预期输出如下所示：

a,d,UUID1, UUID2, UUID3
1,2,2,'add',
3,5,,,23-2-2018

a,d,UUID1, UUID2, UUID3
1,,2,'add',
3,5,,,23-2-2018

与：

---

我还有一个问题，它不允许在列“a”或“d”中使用nan值。相反，它完全删除了该行。