How to concatenate efficiently multiple pandas data frame columns based on conditionals?

Question

I'm trying to concatenate multiple columns in Python. The columns to concatenate vary based on the values of some other columns. How can you do that efficiently?

I've already tried creating a key that groups the conditionals fields and combine that with a for loop that checks each row if it is in a specific group. Of course this takes really long to complete.

For example, given a data frame (df):

df = pd.DataFrame({'cond_1': ['A', 'B', 'B', 'C', 'D'], 
              'cond_2': ['one', 'two', 'three', 'three', 'four'], 
             'concat_1': ['Mon', 'Tue', 'Fri', 'Wed', 'Thu'],
             'concat_2': ['Sep', 'Oct', 'Oct', 'Nov', 'Dec'],
             'concat_3': ['first', 'second', 'second', 'third', 'fourth']})

I have the following set of rules:
- if cond_1 = 'A' then concat_1 + concat_2
- if cond_1 = 'B' then if cond_2 = 'two' then concat_1 + concat_3 else concat_1 + concat_2
- if cond_1 in ('C', 'D') then concat_2 + concat_3

that should result in the following:

cond_1 | cond_2 | concat_1 | concat_2 | concat_3 | result    
---------------------------------------------------------
A      | one    | Mon      | Sep      | first    | MonSep  
B      | two    | Tue      | Oct      | second   | Tuesecond  
B      | three  | Fri      | Oct      | second   | FriOct  
C      | three  | Wed      | Nov      | third    | Novthird  
D      | four   | Thu      | Dec      | fourth   | Decfourth  

Thanks for your help!

Answer 1

You do this with apply using a function to do the if check and concatenation like this

def concate_it(row):
    if row['cond_1'] == 'A':
        return row['concat_1'] + row['concat_2']
    elif row['cond_1'] == 'B' and row['cond_2'] == 'two':
        return row['concat_1'] + row['concat_3']
    elif row['cond_1'] == 'B' and row['cond_2'] != 'two':
        return row['concat_1'] + row['concat_2']
    elif row['cond_1'] in ['C', 'D']:
        return row['concat_2'] + row['concat_3']

df['result'] = df.apply(lambda row : concate_it(row), axis=1)