Pandas - 根据列值有条件地选择列名

Question

如何根据python中DataFrame的列值选择列名？非常感谢，

import pandas as pd

foo = pd.DataFrame([['A',1,2,4],['B',3,4,2],['C',5,6,1]], columns=('a', 'x', 'y','z'))

foo.set_index('a')

Out[1]:

a  x  y  z

A  1  2  4

B  3  4  2

C  5  6  1

对于每一行，检查列值，如果值 <= 2，则记录列名称。

对于上面的例子，输出应该是：

Output[2]: 

a  1  2  3

A  x  y  

B  z

C  z

Answer 1

使用melt，然后过滤值，然后使用值assign的新键cumcountdf=foo.melt('apple').\&nbsp; &nbsp; &nbsp; &nbsp; loc[lambda x : x.value<=2].\&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; assign(value=lambda x : x.groupby('apple').cumcount()+1)df.pivot('apple','value','variable')Out[56]:&nbsp;value&nbsp; 1&nbsp; &nbsp; 2apple&nbsp; &nbsp; &nbsp; &nbsp;&nbsp;A&nbsp; &nbsp; &nbsp; x&nbsp; &nbsp; yB&nbsp; &nbsp; &nbsp; z&nbsp; NaNC&nbsp; &nbsp; &nbsp; z&nbsp; NaN

Answer 2

这是使用列标签mul替换值的一种方法<= 2。然后排序以在每行的末尾放置空字符串。import pandas as pdfoo = pd.DataFrame([['A',1,2,4],['B',3,4,2],['C',5,6,1]], columns=('apple', 'x', 'y','z'))arr = (foo.iloc[:, 1:] <= 2).mul(foo.columns[1:].to_series(), axis=1)arr_sorted = arr.apply(sorted, key=bool, reverse=True, axis=1).valuesfoo.iloc[:, 1:] = pd.DataFrame(arr_sorted.tolist(), columns=foo.columns[1:])print(foo)&nbsp; apple&nbsp; x&nbsp; y z0&nbsp; &nbsp; &nbsp;A&nbsp; x&nbsp; y&nbsp;&nbsp;1&nbsp; &nbsp; &nbsp;B&nbsp; z&nbsp; &nbsp; &nbsp;2&nbsp; &nbsp; &nbsp;C&nbsp; z&nbsp;&nbsp;