You can sum the columns you're interested in, if their type is string:
In [11]: df = pd.read_csv('data.csv', index_col='Customer_ID')
In [12]: df
Out[12]:
Date Time OtherColumns A B C Cost
Customer_ID
1003 January 2:00 Stuff 1 5 2 519
1003 January 2:00 Stuff 1 3 2 530
1003 January 2:00 Stuff 1 3 2 530
1004 Feb 2:00 Stuff 1 1 0 699
In [13]: res = df[list('ABC')].astype(str).sum(1) # cols = list('ABC')
In [14]: res
Out[14]:
Customer_ID
1003 152
1003 132
1003 132
1004 110
dtype: float64
To get the csv, you can first use to_frame
to add the desired column name:
In [15]: res.to_frame(name='ABC') # ''.join(cols)
Out[15]:
ABC
Customer_ID
1003 152
1003 132
1003 132
1004 110
In [16]: res.to_frame(name='ABC').to_csv('new.csv')