How to pythonically return different dataframe column calculations

Question

I have a function which goes something like:

import pandas as pd
import numpy as np


def my_fun(df, calc_type):
    if calc_type == 'sum':
        return df['col_name'].sum()
    elif calc_type == 'mean': 
        return df['col_name'].mean()
    elif calc_type == 'std': 
        return df['col_name'].std()

# sample df
df = pd.DataFrame({'col_name': np.random.normal(25, 3.5, 1000)})

# function call
my_func(df, 'sum')

I was wondering if there's a more elegant, one liner, way to do so. There is getattr which would be great if it worked on methods. Is there something similar?

jezrael · Accepted Answer · 2022-07-07T07:59:43.197

4

IIUC use:

def my_fun(df, calc_type):
      #specify possible functions 
     if calc_type in ['sum','mean','min','max', 'std']:
         return df['col_name'].agg(calc_type)
     else:
         print ('wrong calc_type')

Thank you for comment @Himanshu Poddar , if never calc_type is invalid solution is:

def my_fun(df, calc_type):
    return df['col_name'].agg(calc_type)

EDIT: If need to use getattr use:

out = getattr(s, calc_type)()

edited Jul 07 '22 at 07:59

answered Jul 06 '22 at 05:57

jezrael

822,522
95
1,334
1,252

1

if OP gurantees that `calctype` will never be invalid then one liner should be `return df['col_name'].agg(calc_type)` – Himanshu Poddar Jul 06 '22 at 06:01
It looks great. Is there a method which would work for non Pandas objects? – Eric Johnson Jul 06 '22 at 06:13
@GeorgeIrwin - Not sure if understand, can you explain more? What is `non Pandas objects` ? – jezrael Jul 06 '22 at 06:16
Think even on user made class with a method `my_methodA` and `my_methodB`. How could you apply them? – Eric Johnson Jul 06 '22 at 06:20
@GeorgeIrwin - So it means need custom function/ method like [this](https://pandas.pydata.org/docs/development/extending.html#subclassing-pandas-data-structures) ? – jezrael Jul 06 '22 at 06:31
@jezrael, No, think out of the Pandas eco system. Any object which has a method. We can access the method by `object.method()`. I am asking if there an alternative way like in the case of attributes (`getttr(object, attribute)`). – Eric Johnson Jul 06 '22 at 07:28
@GeorgeIrwin - I understand, so need [this](https://stackoverflow.com/a/48039460/2901002) ? – jezrael Jul 06 '22 at 08:17
@GeorgeIrwin why not use this `print(getattr(s, calc_type)())` – Himanshu Poddar Jul 06 '22 at 09:18
@jezrael, Could you add another answer with what's on the link you posted? – Eric Johnson Jul 07 '22 at 07:56
@GeorgeIrwin - added to answer. – jezrael Jul 07 '22 at 07:59
1

@jezrael, thank you. I marked it and +1. – Eric Johnson Jul 07 '22 at 13:31

How to pythonically return different dataframe column calculations

1 Answers1