Announcement

Collapse
No announcement yet.

TWD data visualization

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • TWD data visualization

    Some TWD visualizations I've made. Most are made possible by the positions of kills/deaths being stored since the 2nd of December 2020.

    Interested to know your opinion on why it shows what it shows.

    Legend:
    white/blue = kill
    red = death

    DD kill/death positions
    Click image for larger version

Name:	javaw_TRODXnomLt.png
Views:	336
Size:	171.9 KB
ID:	1350703

    JD
    Click image for larger version

Name:	javaw_dZrn7tCu87.png
Views:	330
Size:	126.8 KB
ID:	1350704

    Combined match kill/death positions, pairs linked
    Click image for larger version

Name:	javaw_OVWxn74dnA.png
Views:	311
Size:	1.22 MB
ID:	1350705
    Interestingly the distribution shows a definite bias to the bottom right corner. The most likely reason seems to be the position of the radar, which is less in the way when the opponent is in the top left area of the screen. This clear bias made us look at the win percentage of each starting side, which was locked to the top left for the challenging team, and the bottom right for the accepting team.

    When looking at round wins per side for TWD:
    • DD all time: 36289 vs 38664 (48.42% vs 51.58%)
    • DD 2020: 7561 vs 7601 (49.87% vs 50.13%)
    • JD all time: 39090 vs 41167 (48.71% vs 51.29%)
    • JD 2020: 4133 vs 3923 (51.3% vs 48.7%)
    • BD all time: 3124 - 3062 (50.5% vs 49.5%)
    • BD 2020: 1124 vs 1075 (51.11% vs 48.89%)

    Match wins all time:
    • DD: 14248 - 15676 (47.61% vs 52.39%)
    • JD: 15357 - 16677 (47.94% vs 52.06%)
    Even though these differences are caused by many factors, and can't really be put down to the side a team starts on, the side a team starts on is now randomized for round 1, is the opposite for round 2, and random for the tie breaker. For BDs it is always random.


    TWDD range distribution
    R1: 0 - 240px R2: 241 - 640px R3: 641 - 1280px R4: 1281+px
    Click image for larger version

Name:	javaw_CDIV7Mktzn.png
Views:	311
Size:	14.7 KB
ID:	1350706

    TWDTD range distribution (very low sample size)
    Click image for larger version

Name:	javaw_gtP5HJjKgA.png
Views:	311
Size:	10.1 KB
ID:	1350710
    From these very early results it looks like draft matches tend to play out a bit more on the safe side/further away from the opponent, which is to be expected.

    JD kills/deaths without walls
    Click image for larger version

Name:	opera_3pP1QVtcLk.png
Views:	317
Size:	183.5 KB
ID:	1350707
    Looks like the JD map is used pretty much 100% with not that much difference between areas. Only the walls in the top left and right corners show a definite lack of kills/deaths. The bias for fighting on the right side is less pronounced here than it is in DD but is still there.

    BD base - kills without walls
    Click image for larger version

Name:	opera_DhBZySSMsR.png
Views:	308
Size:	201.0 KB
ID:	1350708
    Shows the positions of the lining spots to break cram, as well as cram, top bar, and side bars being popular/effective/concentrated areas to kill from.

    BD base - deaths without walls
    Click image for larger version

Name:	opera_Rk0yrPaCvj.png
Views:	315
Size:	194.5 KB
ID:	1350709
    The death data show the high concentration of deaths in the entrance, as well as higher density in spots where terrs get cornered like the ears, bottom corners, and bottom pockets. BDs also show a slight bias to the right.

  • #2
    This is fking awesome.

    Comment


    • #3
      fkn geek
      4:BigKing> xD
      4:Best> i'm leaving chat
      4:BigKing> what did i do???
      4:Best> told you repeatedly you cannot use that emoji anymore
      4:BigKing> ???? why though
      4:Best> you're 6'4 and black...you can't use emojis like that
      4:BigKing> xD

      Comment


      • #4
        Would be interested in seeing updated base images based on the new map.

        Comment

        Working...
        X