Please provide python code to address the following scenario:

Background In our newsfeed, we show both news content and social updates to our users. In the case of social updates, we want to surface to a user (userA) content about the other users on the site he or she cares most about. We utilize an "affinity score" between userA and other users to quantify the strength of their relationship. In our newsfeed, our goal is then to rank order social updates according to this affinity score.

Prompt You are given a large CSV file of affinity scores for every unique user-user pair. The format of the data is:

Your task is to find, for every user_id value, the other user_id value that maximizes the affinity score. For example, if we had 3 total user_ids, the input data would look something like:

You should provide python code that reads the data from this CSV and returns the desired "best matches" (highest affinity scores). The exact format of your functions output is up to you, but it should be something that is easy to validate and that can be called by other functions. If you make any assumptions about these data - you are welcome to do so - please make these assumptions clear in your solution and write a sentence or two to justify them. Reminder: Your CSV data source is large, so loading the entire thing into memory at once is not the best approach.

UserA UserB Affinity score user.id user jd affinity.score,2 user id1 user_id2 user.id user d affinity.scorei,.3 user_id3 affinity_score1,3 user_idzuser id ffinity_score21 user_idzuser id ffinity_score23 user_idn user idn-2 affinity scorenn-2 user.idn user.idn-1 affinity.scorenn-1