Python Pandas: get columns by values in row

Question

For the purposes of my clustering algorythm I need to iterate over a word/document matrix row by row, and for every row get the submatrix of all columns where this row has a value of 1, (better even, with the exclusion of the row iterated). Say I have a df:

df = pd.DataFrame({'A': '0 1 0 1 0 1 0 1'.split(),
                 'B': '1 1 0 1 0 0 1 0'.split(),
                 'C': '0 0 0 1 0 0 1 0 '.split(),
                 'D': '0 0 1 0 0 0 0 0'.split()})

   w1 w2 w3 w4
0  0  1  0  0
1  1  1  0  0
2  0  0  0  1
3  1  1  1  0

I need the code to return for the first row

   w2
1  1
2  0
3  1

For the second

   w1 w2
0  1  0
2  0  0
3  1  1

and so on. How do I do that? Can't wrap my mind around it using .iloc

Answer 1

IIUC, I print all those steps in case you need them to understand the process

l=np.where(df.eq(1), df.columns, 'nan')
df_list=[]

for y,x in enumerate(l) :
    print(x)
    print(y)
    print(x[x!='nan'])
    print(df.drop(y)[x[x!='nan']])
    df_list.append(df.drop(y)[x[x!='nan']]) #you can store those df in a list 


['nan' 'w2' 'nan' 'nan']
0
['w2']
   w2
1   1
2   0
3   1
['w1' 'w2' 'nan' 'nan']
1
['w1' 'w2']
   w1  w2
0   0   1
2   0   0
3   1   1
['nan' 'nan' 'nan' 'w4']
2
['w4']
   w4
0   0
1   0
3   0
['w1' 'w2' 'w3' 'nan']
3
['w1' 'w2' 'w3']
   w1  w2  w3
0   0   1   0
1   1   1   0
2   0   0   0