Pythonic way to efficient create dictionaries from Pandas

Question

I have a Pandas dataframe that contains columns id, date_created, rank_1, rank_2, rank_3. Below shows 2 rows of the dataframe.

id	date_created	rank_1	rank_2	rank_3
2223	3/3/21 3:26	www.google.com	www.yahoo.com	www.ford.com
1112	2/25/21 1:35	www.autoblog.com	www.motor1.com	www.webull.com

I am trying to assign a new column to this df and call it rank_dict, which will assign number 3 to the rank_1 URL, number 2 to rank_2 URL and number 1 to rank_3 URL. So the ideal result would look like this:

id	date_created	rank_1	rank_2	rank_3	rank_dict
2223	3/3/21 3:26	www.google.com	www.yahoo.com	www.ford.com	{www.google.com:3, www.yahoo.com:2, www.ford.com:1}
1112	2/25/21 1:35	www.autoblog.com	www.motor1.com	www.webull.com	{www.autoblog.com:3, www.motor1.com:2, www.webull.com:1}

I know how to do this if it's not a Pandas df. For example, if I have these key values lists:

keys = ['www.google.com','www.yahoo.com','www.ford.com']

values = [3, 2, 1]

I can do res_dict = dict(zip(keys, values)) to turn it into the dict: {'www.google.com': 3, 'www.yahoo.com': 2, 'www.ford.com': 1}.

But I couldn't figure out an elegant way to perform this dictionary creation in a Pandas df. Could anyone help me?

Answer 1

One way is apply and use enumerate to get the, well, enumeration:

df['rank_dict'] = (df.filter(like='rank_')
                     .apply(lambda x: {v:3-k for k,v in enumerate(x)}, axis=1)
                  )

Output:

     id   date_created             rank_1           rank_2          rank_3                                          rank_dict
0  2223   3/3/21 3:26     www.google.com    www.yahoo.com     www.ford.com  {'www.google.com': 3, 'www.yahoo.com': 2, 'w...'
1  1112  2/25/21 1:35   www.autoblog.com   www.motor1.com   www.webull.com  {'www.autoblog.com': 3, 'www.motor1.com': 2,...'