if 语句是模棱两可的,因为它评估的 Series 可能为空,可能有一些 True 值或 False 值。因此无法评估。见这里:http ://pandas.pydata.org/pandas-docs/dev/gotchas.html#using-if-truth-statements-with-pandas,http : //pandas.pydata.org/pandas-docs/dev /indexing.html#boolean-indexing,http://pandas.pydata.org/pandas-docs/dev/basics.html#flexible-comparisons _ _
这适用于 0.12
In [2]: df = DataFrame(np.arange(20).reshape(10,2),columns=['A','B'])
In [3]: df['status'] = ['A']*4 + ['invalid'] * 2 + ['P'] * 4
在 0.13 中不需要此步骤
In [4]: df['end_date'] = np.nan
从@DSM 编辑
In [5]: df.loc[(df['status'].isin(['A','P'])),'end_date'] = Timestamp('20130101')
In [6]: df
Out[6]:
A B status end_date
0 0 1 A 2013-01-01 00:00:00
1 2 3 A 2013-01-01 00:00:00
2 4 5 A 2013-01-01 00:00:00
3 6 7 A 2013-01-01 00:00:00
4 8 9 invalid NaN
5 10 11 invalid NaN
6 12 13 P 2013-01-01 00:00:00
7 14 15 P 2013-01-01 00:00:00
8 16 17 P 2013-01-01 00:00:00
9 18 19 P 2013-01-01 00:00:00