Melt multiple boolean columns in a single column in pandas

Question

I have a pandas dataframe like this

  Windows Linux Mac
0 True    False False
1 False   True  False
2 False   False True

I want to combine these three columns in a single column like this

  OS
0 Windows
1 Linux
2 Mac

I know that I can write a simple function like this

def aggregate_os(row):
   if row['Windows'] == True:
      return 'Windows'
   if row['Linux'] == True:
      return 'Linux'
   if row['Mac'] == True:
      return 'Mac'

which I can call like this

df['OS'] = df.apply(aggregate_os, axis=1)

The problem is that my dataset is huge and this solution is too slow. Is there a more efficient way of doing this aggregation?

Answer 1

idxmax

df.idxmax(1).to_frame('OS')

        OS
0  Windows
1    Linux
2      Mac

---

np.select

pd.DataFrame(
    {'OS': np.select([*map(df.get, df)], [*df])},
    df.index
)

        OS
0  Windows
1    Linux
2      Mac

---

dot

df.dot(df.columns).to_frame('OS')

        OS
0  Windows
1    Linux
2      Mac

---

np.where

Assuming only one True per row

pd.DataFrame(
   {'OS': df.columns[np.where(df)[1]]},
    df.index
)

        OS
0  Windows
1    Linux
2      Mac

Answer 2

Using boolean indexing with stack and rename

df_new = df.stack()
df_new[df_new].reset_index(level=1).rename(columns={'level_1':'OS'}).drop(columns=0)

Output

        OS
0  Windows
1    Linux
2      Mac