Extracting unique pairs from a data set based on another value

[nilamo]

I'm not sure how it could be done in pandas. I'm also not sure what the match_id does, except act as an auto-increment, but I included it anyway for the fun of it.

Here's how it could be done with a simple loop:

>>> import pandas as pd
>>> data = {'id1': [1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3],
... 'id2': [5, 6, 7, 8, 9, 5, 6, 7, 8, 9, 5, 6, 7, 8, 9],
... 'score': [30, 19, 29, 25, 14, 27, 26, 24, 23, 12, 20, 21, 16, 15, 17]}
>>> df = pd.DataFrame(data)
>>> df = df.sort_values(by="score", ascending=False)
>>> df
    id1  id2  score
0     1    5     30
2     1    7     29
5     2    5     27
6     2    6     26
3     1    8     25
7     2    7     24
8     2    8     23
11    3    6     21
10    3    5     20
1     1    6     19
14    3    9     17
12    3    7     16
13    3    8     15
4     1    9     14
9     2    9     12
>>> matches = {"id1": [], "id2": [], "score": [], "match_id": []}
>>> for row in df.values:
...   id1, id2, score = row
...   if id1 not in matches["id1"] and id2 not in matches["id2"]:
...     matches["id1"].append(id1)
...     matches["id2"].append(id2)
...     matches["score"].append(score)
...     matches["match_id"].append(len(matches["match_id"])+1)
...
>>> matches
{'id1': [1, 2, 3], 'id2': [5, 6, 9], 'score': [30, 26, 17], 'match_id': [1, 2, 3]}
>>> new_df = pd.DataFrame(matches)
>>> new_df
   id1  id2  score  match_id
0    1    5     30         1
1    2    6     26         2
2    3    9     17         3