How to apply multiple criteria condition using apply function-CodePudding

I have a dataframe as below:

    A         B
0   Morning  NaN
1   24       afternoon
2   34       23
3   23       16
4   NaN      13
5   NaN      NaN
6   15.5     1

I want to have a new column 'sumA' which will 2 for these cells (except string and NAN) , so i wrote the code below:

df['sumA'] = df['A'].apply(lambda x: x   2 if (df['A'].notnull) and (type(df['A'])== int) else x )

i got the result below , it doesn't 2 to these cells which i expected. Could you please help assist on this?

     A          B               sumA
0   Morning     NaN              Morning
1   24         afternoon         24
2   34          23               34
3   23          16               23
4   NaN         13               NaN
5   NaN         NaN              NaN
6   15.5        1                15.5

CodePudding user response：

You can convert values to numeric with errors='coerce' for NaNs if not numbers and then add 2, last if need non numeric values use fillna:

df['sumA'] = pd.to_numeric(df['A'], errors='coerce').add(2).fillna(df['A'])
print (df)
         A          B     sumA
0  Morning        NaN  Morning
1       24  afternoon     26.0
2       34         23     36.0
3       23         16     25.0
4      NaN         13      NaN
5      NaN        NaN      NaN
6     15.5          1     17.5

Better if only numeric columns finally - numbers or missing values:

df['sumA'] = pd.to_numeric(df['A'], errors='coerce').add(2)
print (df)
         A          B  sumA
0  Morning        NaN   NaN
1       24  afternoon  26.0
2       34         23  36.0
3       23         16  25.0
4      NaN         13   NaN
5      NaN        NaN   NaN
6     15.5          1  17.5

If need custom function:

#convert to numeric - if strings convert to NaNs
df['A'] = pd.to_numeric(df['A'], errors='coerce')

#pass custom function
df['sumA'] = df['A'].apply(lambda x: x   2 if pd.notna(x) else x )
print (df)
      A          B  sumA
0   NaN        NaN   NaN
1  24.0  afternoon  26.0
2  34.0         23  36.0
3  23.0         16  25.0
4   NaN         13   NaN
5   NaN        NaN   NaN
6  15.5          1  17.5

If really need mixing numeric and non numeric:

#convert strings numbers to numbers (float, int)
df['A'] = pd.to_numeric(df['A'], errors='coerce').fillna(df['A'])

#apply function
df['sumA'] = df['A'].apply(lambda x: x   2 if pd.notna(x) and not isinstance(x, str) else x)
print (df)
         A          B     sumA
0  Morning        NaN  Morning
1     24.0  afternoon     26.0
2     34.0         23     36.0
3     23.0         16     25.0
4      NaN         13      NaN
5      NaN        NaN      NaN
6     15.5          1     17.5