如何删除Pandas系列重复索引的额外副本？

Question

我有一个s重复索引系列:

>>> s
STK_ID  RPT_Date
600809  20061231    demo_str
        20070331    demo_str
        20070630    demo_str
        20070930    demo_str
        20071231    demo_str
        20060331    demo_str
        20060630    demo_str
        20060930    demo_str
        20061231    demo_str
        20070331    demo_str
        20070630    demo_str
Name: STK_Name, Length: 11

我只想通过以下方式保留唯一行和重复行的一个副本:

s[s.index.unique()]

Pandas 0.10.1.dev-f7f7e13 给出以下错误消息

>>> s[s.index.unique()]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "d:\Python27\lib\site-packages\pandas\core\series.py", line 515, in __getitem__
    return self._get_with(key)
  File "d:\Python27\lib\site-packages\pandas\core\series.py", line 558, in _get_with
    return self.reindex(key)
  File "d:\Python27\lib\site-packages\pandas\core\series.py", line 2361, in reindex
    level=level, limit=limit)
  File "d:\Python27\lib\site-packages\pandas\core\index.py", line 2063, in reindex
    limit=limit)
  File "d:\Python27\lib\site-packages\pandas\core\index.py", line 2021, in get_indexer
    raise Exception('Reindexing only valid with uniquely valued Index '
Exception: Reindexing only valid with uniquely valued Index objects
>>> 

那么如何删除额外的重复行系列,保持唯一行和重复行的只有一个副本的有效方式？(一行更好)

Answer 1

您可以通过索引进行分组并应用一个为每个索引组返回一个值的函数.在这里,我采取第一个值:

In [1]: s = Series(range(10), index=[1,2,2,2,5,6,7,7,7,8])

In [2]: s
Out[2]:
1    0
2    1
2    2
2    3
5    4
6    5
7    6
7    7
7    8
8    9

In [3]: s.groupby(s.index).first()
Out[3]:
1    0
2    1
5    4
6    5
7    6
8    9

UPDATE

解决BigBug关于将MultiIndex传递给Series.groupby()时崩溃的评论:

In [1]: s
Out[1]:
STK_ID  RPT_Date
600809  20061231    demo
        20070331    demo
        20070630    demo
        20070331    demo

In [2]: s.reset_index().groupby(s.index.names).first()
Out[2]:
                    0
STK_ID RPT_Date
600809 20061231  demo
       20070331  demo
       20070630  demo

Answer 2

您可以使用duplicated(默认情况下保留第一个值)对数据进行子集化index.使用@ Zelazny7示例:

s = pd.Series(range(10), index=[1,2,2,2,5,6,7,7,7,8])

In [130]: s[~s.index.duplicated()]
Out[130]: 
1    0
2    1
5    4
6    5
7    6
8    9
dtype: int64

Answer 3

一种方法是使用drop和index.get_duplicates:

In [43]: df
Out[43]: 
                      String
STK_ID RPT_Date             
600809 20061231  demo_string
       20070331  demo_string
       20070630  demo_string
       20070930  demo_string
       20071231  demo_string
       20060331  demo_string
       20060630  demo_string
       20060930  demo_string
       20061231  demo_string
       20070331  demo_string
       20070630  demo_string

In [44]: df.drop(df.index.get_duplicates())
Out[44]: 
                      String
STK_ID RPT_Date             
600809 20070930  demo_string
       20071231  demo_string
       20060331  demo_string
       20060630  demo_string
       20060930  demo_string