Fill NaN of DF with the values below

I have a DF like this, it’s a transposed DF (pd.transpose()) :

         0    1    2
COL_5  NaN  NaN  NaN
COL_4  NaN  NaN    4
COL_3  NaN   12    7
COL_2   15    4   11
COL_1    7    8    9

I want to replace the NaN’s with the values below :

         0    1    2
COL_5   15   12    4
COL_4    7    4    7
COL_3         8   11
COL_2              9
COL_1

I don’t know how to do it…

>Solution :

You can use:

out = (df
   .apply(lambda c: pd.Series(c.dropna().to_numpy()))
   .reindex(range(len(df)))
   .set_axis(df.index)
)

Better alternative if you only have the NaNs at the top and no more afterwards:

out = df.apply(lambda c: c.shift(-c.isna().sum()))

output:

          0     1     2
COL_5  15.0  12.0   4.0
COL_4   7.0   4.0   7.0
COL_3   NaN   8.0  11.0
COL_2   NaN   NaN   9.0
COL_1   NaN   NaN   NaN