Ufdjrw

I do know some posts are quite similar to my question but none of them succeded in giving me the correct answer. I want, for each row of a pandas dataframe, to perform the sum of values taken from several columns. As the number of columns tends to vary, I want this sum to be performed from a list of columns.

At the moment my code looks like this:

df['Sum'] = df['Col A'] + df['Col E'] + df['Col Z']

I want it to be something like :

df['Sum'] = sum(list_of_my_columns)

or

df[list_of_my_columns].sum(axis=1)

But both of them return an error. Might be because my list isn't properly created? This is how I did it:

list_of_my_columns = [df['Col A'], df['Col E'], df['Col Z']]

But this doesn't seem to work... Any ideas ? Thank you !

You do not need the list of columns. You need a list of column names:

list_of_my_columns = ['Col A', 'Col E', 'Col Z']
df['Sum'] = df[list_of_my_columns].sum(axis=1)

Let me write as an answer

list_of_my_columns = [df['Col A'], df['Col E'], df['Col Z']]

Using concat

df['Sum']=pd.concat(list_of_my_columns,axis=1 ).sum(axis=1)

All you have to do is create a Series (which is a column in pandas) and take the sum of each row and append it to the Series. Then just add the Series to the dataframe.

import pandas as pd

df = pd.read_csv('input.csv', header=None)
col = pd.Series()

for i, row in df.iterrows():
 col.set_value(i, (row[0] + row[1] + row[2]))

df = df.assign(sum=col)

print(df)