| Forum Index > LoL General |
Starting Page 94 spamming will in GD will be warned, please don't post for the sake of post count. Keep it civil. Please take website feedback to http://www.liquidlegends.net/forum/website-feedback/ | ||
|
Gahlo
United States35166 Posts
On January 06 2015 12:22 GolemMadness wrote: Is 17-0-13 the best mastery page for Eve these days? Is 21/9 even viable? I want to try bringing back some Eve. Depends, how all in do you want to go on a the success of a lvl 3 gank? | ||
|
ShaLLoW[baY]
Canada12499 Posts
On January 06 2015 12:43 Gahlo wrote: Depends, how all in do you want to go on a the success of a lvl 3 gank? 21-9 is fine | ||
|
krndandaman
Mozambique16569 Posts
| ||
|
Sufficiency
Canada23833 Posts
For example the bottom group (Group 4) has most of the ADC champions. This makes sense, since there are probably many "ADC mains" who plays a lot of ADC (thus high correlation for ADC champion usage) and very little everything else. NOTE: I added the horizontal black lines manually for easier reading. I am pretty sure it's not possible to sketch those using off the shelf methods. Here's a version without the black lines. ![[image loading]](http://i.imgur.com/XzGNVno.png) If you like these kinds of graphs you may want to look at this one: it's drawn using the exact same methodology from a Nature paper: http://www.nature.com/srep/2013/130408/srep01583/fig_tab/srep01583_F6.html It's a pretty neat way to group items together, although personally I am pretty meh when it comes to hierarchical clustering. | ||
|
Goumindong
United States3529 Posts
| ||
|
ShaLLoW[baY]
Canada12499 Posts
| ||
|
MidnightGladius
China1214 Posts
I'd be interested to know if there are any champs that are "go-to" for people forced to play a role that's not their main. Conventional wisdom is that you play something like Caitlyn or Ziggs, is there any way to test that? | ||
|
Slusher
United States19143 Posts
| ||
|
Goumindong
United States3529 Posts
In a heiraechical cluster groups are formed in pairwise based on distance*. Those groups (and any odd ones out) are then grouped pairwise again based on distance. This is repeated until there is only one group at the top. The lines on the side are a visual representation of that process. The graph is organized in a way to make the roots obvious (ie no crossing lines) which also has the nice effect of clustering the high correlations around the diagonal. *which can have many definitions | ||
|
M2
Bulgaria4134 Posts
On January 06 2015 15:41 ShaLLoW[baY] wrote: does anyone else think reksai top is just absolutely broken? I don't think there is a single champion who wins trades with reksai top, perhaps some ranged champs if they are able to trade without taking damage, but hardly. The only consolidation is that around lvl 12-13 if she did not rekt to oblivion, many top laners catch up with her since she has only one damage ability. | ||
|
GhandiEAGLE
United States20754 Posts
| ||
|
zer0das
United States8519 Posts
| ||
|
M2
Bulgaria4134 Posts
On January 06 2015 17:25 zer0das wrote: Dunno, I played against Rek'sai as Jax and it kinda sucked around 6-8, but once I got over the hump I just mashed her face. Granted its Jax... This is her only weakness, she does not do much more after lvl 9, leveling W/E/R do not add a lot for Reksai | ||
|
ShaLLoW[baY]
Canada12499 Posts
and now i go back into queue and he's on my team again but picks fiora flash ignite, why would you be such a cunt for one game and then play properly? assuming he isn't about to troll again nope he switched to tp clarity fuck this game man | ||
|
I_Love_Bacon
United States5765 Posts
| ||
|
Baffels
United States1486 Posts
http://www.azubu.tv/ngtv | ||
|
Sufficiency
Canada23833 Posts
On January 06 2015 15:44 MidnightGladius wrote: Thanks for collecting and organizing the data. I'm afraid that I'm not familiar with the methodology - does each branch of the tree on the left side of the graph correspond to an equally large correlation? In other words, what's the relationship between champions in sets {2,4,6} and {1,3,5,7}? I'd be interested to know if there are any champs that are "go-to" for people forced to play a role that's not their main. Conventional wisdom is that you play something like Caitlyn or Ziggs, is there any way to test that? My bad. The heatmap itself is just a giant correlation matrix. This is why the diagonals are red... since the diagonals = 1. The groups often show up as white-ish due to 0.4ish correlation. As for "go-to champion"... maybe? I need to capture people who are a certain kind of "main" (say 50% of the game played is adc) then look for what else they play. | ||
|
M2
Bulgaria4134 Posts
For what I understand you can swap (tier 2) jungle items for free now. Need sustain go purple, need to gank take blue :-) looks a little weird to me. And finally Sona nerf, this overwhelming high mobility, high health, high outplay potential and no flash team fight dependent monster has been nerfed... | ||
|
Scip
Czech Republic11293 Posts
On January 06 2015 22:09 M2 wrote: http://www.surrenderat20.net/2015/01/15-pbe-update.html For what I understand you can swap (tier 2) jungle items for free now. Need sustain go purple, need to gank take blue :-) looks a little weird to me. And finally Sona nerf, this overwhelming high mobility, high health, high outplay potential and no flash team fight dependent monster has been nerfed... noo, nee-chan! What did they do to your face.. t.t | ||
|
Alaric
France45622 Posts
| ||
| ||
StarCraft 2 StarCraft: Brood War Dota 2 League of Legends Super Smash Bros Heroes of the Storm Other Games |
|
Replay Cast
RongYI Cup
Maru vs Cyan
Solar vs Krystianer
uThermal 2v2 Circuit
BSL 21
Replay Cast
Wardi Open
Monday Night Weeklies
OSC
Replay Cast
WardiTV Invitational
[ Show More ] |
|
|
EPT
