General Discussion

General Discussionhow to get dota 2 data?

how to get dota 2 data? in General Discussion
BlackXargon

    This is me wanting to converge the 2 things that I spend the most time on:
    1) working as a Solutions Architect for a leading Analytics (Data Warehouse/BI/Big Data) company
    2) Playing Dota 2

    What I want?
    1) dota 2 match data - mouse clicks, hero positions etc..
    specs:
    - sample data for a specific region
    - previous and current patch
    - not sure about how big the size is but my demo environment can probably ingest about ~50GB of data

    If anyone has tried this before, i have a few questions:
    1) is it API-based?
    2) technically, can i parse the match data and get any info about the match?
    3) has anyone already identified patterns on how to parse the data?

    What I want to do?
    1) publish results based on suggestions/use cases from dotabuff community

    3.14159265358979

      Dota 2 API only provides basic match stats like who won, who played, what’s their KDA. In order to get more data you will need to parse replay files - this is what Dotabuff Truesight does: https://secure.dotabuff.com/truesight

      BlackXargon

        ^ how to get detailed data like TrueSight?

        Miku Plays

          im pretty sure you need to talk to the dotabuff admins about this

          BlackXargon

            I'm guessing they won't share it to me because it's like telling someone their "trade secret"

            CAN'T WIN

              ^
              TRY IT!

              Lawliepop

                Hi there,

                We don't do custom data sets because we can't afford to spend developer/server time on requests, and our data is not really built for it. If you want more detailed, truesight data you can always try parsing a few replays on your own. Our parser is open source: https://github.com/dotabuff/yasha
                For less detailed match data I recommend just using Valve's APi.

                BlackXargon

                  ^ I didn't thought that they have this in GitHub,.. this should be cool! Thanks!

                  Lawliepop

                    np, excited to read your results!

                    QQ

                      I am profane of programming, i remember this other project to parse replay, don't know how it work or if it's still running.
                      https://github.com/skadistats/skadi

                      Кошеня

                        just want to say big respect whoever build turesight.

                        CaptainDreads

                          Not quite what you were looking for but it would be nice to extract these stats:

                          - Hero damage breakdown per skill, item & autoattacks
                          - Total enemy stun time per hero
                          - Total enemy slow time per hero

                          These stats are printed into the console at the end of the game so I presume they can be extracted from the replay somehow. I've not looked at the format so I couldn't necessarily tell how easy it would be to do.

                          BlackXargon

                            @Lawliepop how do I get the raw data? Is it this protobuf dota2demo thing I need to work on?

                            Basically, I just want to work on raw data sets so I can transfer them to my environment..
                            At the moment, I'm not really interested in building a full-fledged parsing program

                            I just want to run some pattern matching algorithms on the mouse clicks :-/

                            BlackXargon

                              bump :-/