我阅读了下面的DataFrame,其中有许多与S&PCOMP相似的列,在列名的末尾带有(PO)和(PI)尾部。
Date S&PCOMP(PO) S&PCOMP(PI) NASA100(PO) NASA100(PI)
0 1978-09-13 00:00:00 nan 106.34 someValue someValue
1 1978-09-14 00:00:00 nan 105.10000000000001 someValue someValue
2 1978-09-15 00:00:00 nan 104.12
3 1978-09-18 00:00:00 nan 103.21000000000001
4 1978-09-19 00:00:00 nan 102.53
5 1978-09-20 00:00:00 nan 101.73
6 1978-09-21 00:00:00 nan 101.9
7 1978-09-22 00:00:00 nan 101.84
8 1978-09-25 00:00:00 nan 101.86
9 1978-09-26 00:00:00 nan 102.62
10 1978-09-27 00:00:00 nan 101.66
我想使用正则表达式和多索引针对每个单个列名将其重组为以下DataFrame。本质上,我将PO和PI值用作2列,并使用列名称库作为索引垂直扩展DF。如您所见,日期滚动……意味着对于每个唯一的列名称库,我都有相同的天数。
Date Open Close
S&PCOMP 1978-09-13 00:00:00 nan 106.34
S&PCOMP 1978-09-14 00:00:00 nan 105.10000000000001
S&PCOMP 1978-09-15 00:00:00 nan 104.12
S&PCOMP 1978-09-18 00:00:00 nan 103.21000000000001
S&PCOMP 1978-09-19 00:00:00 nan 102.53
S&PCOMP 1978-09-20 00:00:00 nan 101.73
S&PCOMP 1978-09-21 00:00:00 nan 101.9
S&PCOMP 1978-09-22 00:00:00 nan 101.84
S&PCOMP 1978-09-25 00:00:00 nan 101.86
S&PCOMP 1978-09-26 00:00:00 nan 102.62
S&PCOMP 1978-09-27 00:00:00 nan 101.66
NASA100 1978-09-13 00:00:00 someValue someValue
NASA100 1978-09-14 00:00:00 someValue someValue
用大熊猫实现这一目标的最简单方法是什么?我可以使用正则表达式吗?