熊猫将0/1数据框条目映射到列名

Question

I have a data frame in pandas with entries that are either 0 or 1. I would like to reduce this to a single list of strings that are the result from concatentating column names wherever there is a 1.

For a toy example suppose my data frame is

   V1 V2 V3
   0  1  1
   1  1  0
   0  0  0

I would like to have a final result that looks like

"V2,V3"
"V1,V2"
""

I had initially tried using something along the lines of

my_df.apply(lambda x: colnames[x])

thinking it would behave similarly to how numpy handles boolean indices. But, did not achieve what I wanted to do. How should I best accomplish this?

Answer 1

convert the dtype of the df to a bool , then call apply and use the boolean mask to mask the columns, you need to pass param axis=1 to apply the column mask row-wise:

In [47]:
df.astype(bool).apply(lambda x: ','.join(df.columns[x]), axis=1)

Out[47]:
0    V2,V3
1    V1,V2
2         
dtype: object

Your code my_df.apply(lambda x: colnames[x]) won't work because firstly when calling apply on a df without specifying the axis will call the lambda on each column in turn, secondly the 1/0 will interpret this as an index value rather than a boolean flag.