Page 3 of 5 FirstFirst 1 2 3 4 5 LastLast
Results 21 to 30 of 45

Thread: Let there be data!

  1. #21
    Basic Member
    Join Date
    Feb 2012
    Posts
    57
    Playing around with this some more and I've come across a potential issue. 478 is a match in the match database, but if I try to filter match_players by match_id = 478 I get no results. 477 and 479 work fine. Is this a peculiarity specific to that match?

    Related, is there a reason why there are entries in accounts that have no matches in match_players?

  2. #22
    Basic Member jimmydorry's Avatar
    Join Date
    Dec 2012
    Posts
    814
    I wanted to use this... but why Postgre?

    I can't find a tool that will let me convert to mysql... let alone view without installing Postgre.

  3. #23
    Basic Member
    Join Date
    Nov 2012
    Posts
    35
    Quote Originally Posted by Phantasmal View Post
    Playing around with this some more and I've come across a potential issue. 478 is a match in the match database, but if I try to filter match_players by match_id = 478 I get no results. 477 and 479 work fine. Is this a peculiarity specific to that match?

    Related, is there a reason why there are entries in accounts that have no matches in match_players?
    Currently home for the holidays so investigation will be difficult before the new year, but extra accounts will probably have matches at a later date which will be in future dumps, and the missing matchplayers is... weird. It might be due to the early change Valve made on private games; it used to be possible to access the data but they changed it after a while for privacy reasons, so it might be that they left the match details available and hid the player details. If that is the case all of those games will be private and none of them should occur past some point in early 2012.

    Quote Originally Posted by jimmydorry View Post
    I wanted to use this... but why Postgre?
    I can't find a tool that will let me convert to mysql... let alone view without installing Postgre.
    To quote a few posts back:
    Quote Originally Posted by Sproinknet View Post
    One of the reasons we didn't go for MySQL is that there are some unicode gotchas, which would be a big issue for Dota 2 where people just love to put all sorts of weird symbols in their names.
    To expand on that, we consider the choice of database to be academic; once the data is available I'm sure someone else who's a MySQL person will be willing to convert and reupload if there is demand. If we had chosen MySQL there would be nothing to stop someone else posting in here asking why we didn't use postgres, so we're not going to be able to please everyone no matter what we do! Someone else has doubtless worked out how to do the conversion and if said person wants to PM a guide on what they learned doing it we'll gladly add it to the dump to save other people some legwork.

    Currently we have around another 2 months ready to upload, but will be AFK until the new year so sadly at least another week or so before we'll be ready to give you complete 2012 data. Once that's done we'll work on a way of letting people get live data in a more convenient format.

  4. #24
    Basic Member
    Join Date
    Mar 2012
    Posts
    51
    this is my old database of the first 15 million matches dota2. + simple Web intf.
    10-15m matches without player names.

    Size: Live MySQL 9gb;
    dump 14gb;
    rar 2gb.

    MySQL 5.5.24.

    DOWNLOAD GDrive | Mirror

    http://i.imgur.com/43usm.png




  5. #25
    Basic Member
    Join Date
    Dec 2011
    Posts
    83
    @natj thanks looks good, going to play with it. Planning on releasing more (regular)?
    edit: seems it goes up to id 15000000 but there are 6,130,168 matches in there

    @Sproinknet Multiple formats would be great, I think it would save a lot of trouble for users. A postgre dump could also work, if it's plain txt, so it can be converted with a script.
    Last edited by Arie; 12-26-2012 at 11:43 AM.

  6. #26
    How are you able to access this information when the API has been removed from the game by the developers? (source: the sticky in this same forum)

  7. #27
    Basic Member
    Join Date
    Dec 2011
    Posts
    83
    Quote Originally Posted by complaintant View Post
    How are you able to access this information when the API has been removed from the game by the developers? (source: the sticky in this same forum)
    Look at the opening post, your question and answer is right in there.

  8. #28
    Basic Member
    Join Date
    Feb 2012
    Posts
    57
    I've been going through the database and I found that the earlier entry with a skill level of 1 or 2 takes place on 2011-11-23 at 13:00:47. Is it safe to assume that all entries after this point with a skill level of 0 are actually normal level games? There's only 3 values for the category for four possibilities -Very High, High, Normal, and Unknown- so I'm trying to find a foolproof way of distinguishing between the last 2.

  9. #29
    Basic Member
    Join Date
    Nov 2012
    Posts
    35
    Quote Originally Posted by Phantasmal View Post
    There's only 3 values for the category for four possibilities -Very High, High, Normal, and Unknown- so I'm trying to find a foolproof way of distinguishing between the last 2.
    That's not quite right - there should be 4 values (0 for unknown, 1/2/3 for normal/high/v high). There aren't very many v high games we have the details for in this dump, but if you look at the following matches you should get one of each:

    • (0) Unknown - 802948
    • (1) Normal - 802934
    • (2) High - 803232
    • (3) Very High - 808602

    If you run something like:
    Code:
    select skill, count(*) from dota2_match_dump.matches group by skill
    you should see that the vast majority of the data will be in the unknown category. We'll be working on getting that missing data when we can, but it isn't parcelled with the game data so we've put it as a lesser priority for now and included what data we do have.

  10. #30
    Basic Member
    Join Date
    Feb 2012
    Posts
    57
    Oh, I had somehow managed to miss 3 entirely. Must have been a dumb error on my part.

    That's unfortunate. I had hoped there would be a small stretch of time where the skill information would be complete but that looks unlikely. Was the Very High data that you have from a single API call? It appears to have precisely 25 matches, but it seems implausible that only 2 very high matches were played between December 3rd and December 26th.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •