Converting pandas dataframe into a dataframe for network analysis

user_7484093 0
Viewed 37

I read multiple questions, but do not see a solution or missing something

I have a dataframe

Author_id Article_id Rank
    100     10         1
    101     10         2
    102     10         3
    100     11         1
    105     11         2
    106     11         3

Expected output:

 Author_id1 Author_id2 Article_id Rank
   100        101         10        1
   100        102         10        1
   100        105         11        1
   100        106         11        1
   101        100         10        2
   101        102         10        2
   102        100         10        3
   102        101         10        3
   105        100         11        2
   105        106         11        2
   106        100         11        3
   106        105         11        3

If my understanding of the the desired dataset is not correct, please advise on how to restructure this data. Thanks!

pandas networking python
1 Answers

If the DataFrame isn't duplicated on ['Author_id', 'Article_id'], you can do a self-merge within article, and filter out the rows that merged with themself.

df1 = (df.merge(df.drop(columns='Rank'), on='Article_id', suffixes = ['1', '2'])
         .query('Author_id1 != Author_id2')
         .sort_values(['Author_id1', 'Rank'])
         .reset_index(drop=True))

    Author_id1  Article_id  Rank  Author_id2
0          100          10     1         101
1          100          10     1         102
2          100          11     1         105
3          100          11     1         106
4          101          10     2         100
5          101          10     2         102
6          102          10     3         100
7          102          10     3         101
8          105          11     2         100
9          105          11     2         106
10         106          11     3         100
11         106          11     3         105
user_43333590
Related