I have the following dataframe in:
a
1 3
2 2
3 Nan
4 3
5 Nan
I need to recode this column so it looks like this:
df_miss_a
1 0
2 0
3 1
4 0
5 1
I've tried:
df_miss_a = np.where(df['a'] == 'Nan', 1, 0)
and
df_miss_a = np.where(df['a'] == Nan, 1, 0)
The above outputs only 0s.
The format of the output is unimportant.
If you have NaNs in your column you can use pd.Series.isna():
df_miss_a = df["a"].isna().astype(int)
print(df_miss_a)
Prints:
1 0
2 0
3 1
4 0
5 1
Name: a, dtype: int64
pd.Series.fillna(0) is simpler and more obvious
pd.Series.fillna(0) would not work: (1) It sets NaN to 0 rather than 1, as required by OP. Of course, this could easily be mitigated by pd.Series.fillna(1); however, this would leave the 2nd problem: (2) OP wants all values that are not NaN to be set to 0. Your solution with fillna() would simply leave them alone, so you would require an extra step, while Andrej Kesely's answer solves both issues at once.
replaceorfillna) only solve part of OP's problem; namely replacing all values that are NaN. However, OP also wants to replace all values that are not NaN.