Matchmaking by TV: What could have been?
  • I guess you expected another one of the countless matchmaking complaining threads, when you clicked on this one. However, you are wrong: I actually try to answer the question: What would have been?

    Since we have some very big data sets available, we can actually investigate this. So what I roughly did was following:

    1. Modrek provided me with the matchup data from CCL and COL
    2. I used this data set to identify the pools, which were used to create the matchups, which were played
    3. I created different matchups according to different matchmaking score (by TV, etc.)
    4. I evaluated the resulting matchups according to different metrics

    Short answer:

    Long answer: I wrote a documentation about the stuff I did, you can find it here. Even if you are not interested in this stuff, I would hint you to look at page 2, since it shows some interesting statistics about the pooling sizes.

    (note: I also posted this in the steam forums)

    last edited by Arne
  • I intended to do something like this too.
    I just skim-read it. It is interesting but you should definitely post a link to your script / analysis on github.

  • Community Manager

    Interesting topic and analysis. Glad to read the results as well!

  • @Loddfafnir:
    I think I won't put the code on github for now, since it may not be easy to understand. There are like 8 scripts involved in it:

    • Some (preprocessing/identification of pools), I already wrote month ago
    • Some steps of the optimization got rather, lengthy, which is why, I put each step of the optimization into a different script as well

    However, if you want to do something similar, feel free to contact me here for comparison of the result or code fragments, which can do certain stuff.

    What do you want to use for programming? (I am using R)

    last edited by Arne
  • This post is deleted!

  • 5
    Posts
  • 413
    Views
  • Log in to reply
  • Looks like your connection to Matchmaking by TV: What could have been? was lost, please wait while we try to reconnect.