在 Pandas DF 行中查找最小日期并创建新列
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/18134794/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Finding the min date in a Pandas DF row and create new Column
提问by dartdog
I have a table with a number of dates (some dates will be NaN) and I need to find the oldest date so a row may have DATE_MODIFIED, WITHDRAWN_DATE, SOLD_DATE, STATUS_DATE etc..
我有一个包含多个日期的表（有些日期是 NaN），我需要找到最旧的日期，所以一行可能有 DATE_MODIFIED、WITHDRAWN_DATE、SOLD_DATE、STATUS_DATE 等。
So for each row there will be a date in one or more of the fields I want to find the oldest of those and make a new column in the dataframe.
因此，对于每一行，我想找到其中最旧的一个或多个字段中的日期，并在数据框中创建一个新列。
Something like this, if I just do one , eg DATE MODIFIED I get a result but when I add the second as below
像这样的事情，如果我只做一个，例如 DATE MODIFIED 我会得到一个结果但是当我添加第二个时，如下所示
table['END_DATE']=min([table['DATE_MODIFIED']],[table['SOLD_DATE']])
I get:
我得到：
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
For that matter will this construct work to find the min date, assuming I create correct date columns initially?
就此而言，假设我最初创建正确的日期列，此构造是否可以找到最小日期？
回答by Viktor Kerkez
Just apply the minfunction along the axis=1.
只需min沿轴 = 1应用该函数。
In [1]: import pandas as pd
In [2]: df = pd.read_csv('test.cvs', parse_dates=['d1', 'd2', 'd3'])
In [3]: df.ix[2, 'd1'] = None
In [4]: df.ix[1, 'd2'] = None
In [5]: df.ix[4, 'd3'] = None
In [6]: df
Out[6]:
d1 d2 d3
0 2013-02-07 00:00:00 2013-03-08 00:00:00 2013-05-21 00:00:00
1 2013-02-07 00:00:00 NaT 2013-05-21 00:00:00
2 NaT 2013-03-02 00:00:00 2013-05-21 00:00:00
3 2013-02-04 00:00:00 2013-03-08 00:00:00 2013-01-04 00:00:00
4 2013-02-01 00:00:00 2013-03-06 00:00:00 NaT
In [7]: df.min(axis=1)
Out[7]:
0 2013-02-07 00:00:00
1 2013-02-07 00:00:00
2 2013-03-02 00:00:00
3 2013-01-04 00:00:00
4 2013-02-01 00:00:00
dtype: datetime64[ns]
回答by Felix Zumstein
If tableis your DataFrame, then use its minmethod on the relevant columns:
如果table是您的 DataFrame，则min在相关列上使用其方法：
table['END_DATE'] = table[['DATE_MODIFIED','SOLD_DATE']].min(axis=1)
回答by fccoelho
A slight variation over Felix Zumstein's
与 Felix Zumstein 的略有不同
table['END_DATE'] = table[['DATE_MODIFIED','SOLD_DATE']].min(axis=1).astype('datetime64[ns]')
The astype('datetime64[ns]')is necessary in the current version of pandas (july 2015) to avoid getting a float64representation of the dates.
astype('datetime64[ns]')在当前版本的Pandas（2015 年 7 月）中，这是必要的，以避免获得float64日期的表示。
