CODEWITHSUNDEEP
pythonpandas
Konstantin
Konstantin

Reputation: 3159

Why the replace method of pandas does not work with None as value?

Let's say we have a DataFrame:

import pandas as pd
df = pd.DataFrame({"a": [1,"NAN",3], "b": ["NAN",5,"NAN"], "c": ["NAN",8,9]})

     a    b    c
0    1  NAN  NAN
1  NAN    5    8
2    3  NAN    9

When I try to preform a regular replace I get incorrect results:

df = df.replace('NAN', None)

   a    b    c
0  1  NAN  NAN
1  1    5    8
2  3    5    9

However, when I have the values as lists it works as expected:

df = df.replace(['NAN'], [None])

    a    b    c
0  1.0  NaN  NaN
1  NaN  5.0  8.0
2  3.0  NaN  9.0

Can you please explain why the behavior is different in these two cases?

PS: pandas version is 1.0.3.

Upvotes: 0

Views: 332

Answers (1)

Daniel Wyatt
Daniel Wyatt

Reputation: 1151

As was answered here:

df.replace("NAN", None) is equivalent to df.replace("NAN", None, method='pad') if you look into the source code.

Both df.replace({"NAN": None}) and df.replace("NAN", np.nan) give you what you want

Upvotes: 2

Related Questions