lambda each row in a column from dataframe that is not 'nan'-CodePudding

In the below code how can I iterate the lambda for rows that are not "nan", "null" or in other words blank?

The output still appends ".pdf" to even the blank rows.

    model1      model2    model3     
 0  File1.pdf   File3.pdf File2.pdf
 1  File2.pdf   nan.pdf   File2.pdf
 2  File3.pdf   nan.pdf   File2.pdf

model_types = []
for col in df.columns:
    model_types.append(col)
    df[col] = df[col].apply(lambda x: f"{x}.pdf" if x != 'nan' else x)  # this part

CodePudding user response：

If you have real NaNs you should be able to add '.pdf' directly in a vectorized manner and this shouldn't impact the NaNs.

model_types = []
for col in df.columns:
    model_types.append(col)
    df[col]  = '.pdf'

Alternatively, you can also use:

model_types = list(df)
df = df.stack().add('.pdf').unstack()
# or
df = df.apply(lambda s: s '.pdf')

Output:

      model1     model2     model3
0  File1.pdf  File3.pdf  File2.pdf
1  File2.pdf        NaN  File2.pdf
2  File3.pdf        NaN  File2.pdf

CodePudding user response：

First convert your nans to None.

df1 = df.where(pd.notnull(df), None)
for col in df1:
    df1[col].apply(lambda x: f"{x}.pdf" if x else None)

CodePudding user response：

For people who want a more "universal" solution which let you deal with:

0 and integers
floats numbers
pd.NA = < NA >
np.nan = NaN

Creating DF:

import pandas as pd
import numpy as np

df = pd.DataFrame({'Col1':[0,"File2",1.23], 'Col2':["File1",pd.NA,"File3"], 'Col3':["File1","File2",np.nan]})
df

index	Col1	Col2	Col3
0	0	File1	File1
1	File2	<NA>	File2
2	1.23	File3	NaN

df1 = df.applymap(lambda x: x '.pdf' if type(x) == str else x, na_action='ignore')
df1

index	Col1	Col2	Col3
0	0	File1.pdf	File1.pdf
1	File2.pdf	<NA>	File2.pdf
2	1.23	File3.pdf	NaN