I have a dataframe df created as,

df = pd.DataFrame([
     [0, 1, 0],
     [1, 0, 0],
     [0, 1, 0]],
     columns=['Mon','Tue','Wed'])

df:
    Mon Tue Wed
0   0   1   0
1   1   0   0
2   0   1   0

For this dataframe, I want to replace the element '1'- to the respective column name and '0' to be replaced with- "", and finally combine all of them.

1 and 0 can also be characters- 'X' and '-'.

Desired Output:
    Day
0   Tue
1   Mon
2   Tue

CodePudding user response：

without duplicate values (i.e. only one 1 per row)

Use idxmax on axis=1:

df.idxmax(axis=1)

NB. if you have '-'/'X' as values, use df.eq('X').idxmax(axis=1)

output:

0    Tue
1    Mon
2    Tue

As dataframe:

df.idxmax(axis=1).rename('Day').to_frame()

output:

   Day
0  Tue
1  Mon
2  Tue

potential duplicates

You can use melt:

df = pd.DataFrame([
     [0, 1, 1],
     [1, 0, 0],
     [0, 1, 0]],
     columns=['Mon','Tue','Wed'])

# below use 'value == "X"' if '-'/'X' values
df.melt(var_name='Day', ignore_index=False).query('value == 1')[['Day']]

output:

   Day
1  Mon
0  Tue
2  Tue
0  Wed

CodePudding user response：

Use DataFrame.dot with columns names separated by comma if possible duplicated 1 per rows:

df1 = df.dot(df.columns   ',').str[:-1].to_frame('Day')
print (df1)
   Day
0  Tue
1  Mon
2  Tue

If multiple values:

df = pd.DataFrame([
     [0, 1, 1],
     [1, 0, 0],
     [0, 1, 0]],
     columns=['Mon','Tue','Wed'])


print (df)
   Mon  Tue  Wed
0    0    1    1
1    1    0    0
2    0    1    0


df1 = df.dot(df.columns   ',').str[:-1].to_frame('Day')
print (df1)
       Day
0  Tue,Wed
1      Mon
2      Tue