Pandas opposite of fillna(0)

Question

Whereas df.fillna(0) fills all NA/NaN values with 0, is there a function to replace all non-NA/NaN values with another value, such as 1?

If the values in my DataFrame are variable-length lists then:

• df.replace() requires that the lists are the same length
• boolean index like df[len(df) > 0] = 1 throws ValueError: cannot insert True, already exists
• pandas.get_dummies() throws TypeError: unhashable type: 'list'

Is there a more straightforward solution?

Answer 1

You could use indexing/assignment with df[df.notnull()] = 1. For instance:

>>> df = pd.DataFrame([[np.nan, 2, 5], [2, 5, np.nan], [2, 5, np.nan]])
>>> df # example frame
    0  1   2
0 NaN  2   5
1   2  5 NaN
2   2  5 NaN

>>> df[df.notnull()] = 1
>>> df
    0  1   2
0 NaN  1   1
1   1  1 NaN
2   1  1 NaN

Answer 2

I don't know of a built-in function, but this works:

import pandas as pd
import numpy as np

df = pd.DataFrame(data={'a':[np.nan, 13, 32]})

 >>    a
   0  NaN
   1   13
   2   32

df = df.applymap(lambda x: 1 if not np.isnan(x) else x)

 >>     a
    0 NaN
    1   1
    2   1