|
![[image loading]](http://i.imgur.com/GdLFvQW.jpg)
Graph based on 114k recorded games in the pro and semi-pro levels. Each node is a player (about 4000 of them). Each edge represents players who have played a game with each other. It's thicker if they have played more games.
The modularity algorithm identified four major components, corresponding to the EU, NA, KR and SEA/CN/TW scenes. (That would be the colouring. The location is independent of it.)
EU is the largest (37%), then comes NA (29%), KR (21%) and SEA++ (12%).
Within each group there are also some subgroups that can be discerned. Most prominently, the SEA, China and Taiwan scenes. In Europe, there is a small group of players from the British Isles and another group of Scandinavians.
The KR and NA scenes appear more homogenized, except for a small group corresponding to a South American scene (mostly Brazil and Chile).
Interestingly, Sen has his own location somewhere in the middle.
The graph diameter is 9, meaning you can skip between any two players in at most 9 degrees of separation (though on average you need just 3.5).
Any experts on gephi here? Maybe you can do some wizardry on this graph.
Edit: I thought the image was supposed to be resized automatically?
Edit 2: Oh, now it is.
Edit 3: Sorry for the JPEG, I think it was PNG when I uploaded.
|
Beautiful! EU > NA? Didn't expect that.
Edit: Actually, thinking about it, it makes a lot of sense + Sen in limbo is funny.
|
On February 02 2013 08:13 TheBB wrote:
Interestingly, Sen has his own location somewhere in the middle.
LOL thats awesome
cool graph btw. this is awesome
|
Neat graph, TheBB!
Wonder where Sen's own little island could be located..
|
Haha, Sen gets his own blob. Sad that the UK's scene is still nowhere near significant enough. We need to step up.
|
On February 02 2013 09:24 Larkin wrote:
Haha, Sen gets his own blob. Sad that the UK's scene is still nowhere near significant enough. We need to step up.
The UK blob consists mostly of the not-so-significant players. People like Bling and Demuslim are squarely in the middle of the EU blob instead.
Jinro, for example, is in KR.
Also, turns out that Goody is the player who has played the largest number of different opponents (352). Who's up for giving out Goody numbers?
|
dude fix you geographical knowledge!!!
joking still awesome that way. 
|
Now put players on a real map based on their hometowns 
Just kidding, this is excellent work and really shines some light on the relations between the servers and their players
|
Amusing to see Sen the anomaly
|
On February 02 2013 09:27 TheBB wrote:
Also, turns out that Goody is the player who has played the largest number of different opponents (352). Who's up for giving out Goody numbers?
You!
You should totally add a Goody number generator in aligulac. :D
|
That's pretty dang cool!
I see that the Korean scene has some kind of "gradual split up" going downwards... Would that be the KeSPA players or something like that, or KeSPA and eSF are not distinguisheable?
|
On February 02 2013 22:44 Epamynondas wrote:
I see that the Korean scene has some kind of "gradual split up" going downwards... Would that be the KeSPA players or something like that, or KeSPA and eSF are not distinguisheable?
I can't find any separation between Kespa and ESF. The split you are referring to looks strange to me too. Maybe it includes Koreans who play frequently overseas, or overseas players who play frequently in Korea? Notables in that place are Jinro, Scarlett and Hyun.
|
First I thought, "This is crap", then I saw Sen, then I thought, "Correct, no problem here".
|
I think Baller himself wouldn't be ashamed of such a graph  Nicely done.
|
I thought there was only 1 Brazilian semipro player, which is actually a full time medic.
|
I was wondering if I could do something like this with players practice partners. You see players mention them at the end of interviews and I thought that might give us some insight into social aspects of progamers (who's the "cool kid" and who doesn't play with others etc).
As for regional comparison, I'm actually surprised that there is a clear distinction between NA and EU. I thought they would almost overlap. I guess you never know until you actually crunch those numbers.
|
Oh hey someone else who uses Gephi.
If you're at all interested in exporting an interactive web version of this, I'd suggest you export your finished graph in gexf format and read it using sigmajs: http://sigmajs.org
A little tweaking and it can be really pretty!
EDIT: just a sample graph I did awhile ago (large gexf file, so may take awhile to load): http://llanim.us/etiNetworkGraph.php?type=&year=2012
Mouse wheel to scroll, hover over a node to hide all but its immediate neighbors.
|
So I suck at graphs, can anyone explain it for people who know less about how this works than the people who posted until now?
What does it mean if nodes are closer together for one scene than for others, what does the colouring of the lines indicate, etc.
|
Awesome graph
I do understand the connections, but the locations are a bit confusing to me.
Does Sen being in the middle means he plays equal amounts vs EU, Kor, Sea and US players?
|
On February 03 2013 05:11 StarVe wrote:
So I suck at graphs, can anyone explain it for people who know less about how this works than the people who posted until now?
What does it mean if nodes are closer together for one scene than for others, what does the colouring of the lines indicate, etc.
So the clustering will group together players (dots) based on who played the most with each other. So a team that plays with members of its own team exclusively would be an isolated group of dots grouped close together.
So each blob (cluster) represents a sub-community in the SC2 community. We notice all the normal communities we talk about (KR, NA, EU, etc.) and this graph helps us see how they relate and some funny things (like Sen doesn't cluster well with KR or NA so he is put in his own cluster in the middle).
The thickness of the lines from one community to another shows how much cross community play there is or how isolated the communities are from one another. Since most of the lines are thick this tells us global play is pretty common, and our communities are not very isolated.
Hope this helps the graph noobs!
|
|
|
|
|
|
EPT
