EPT[Patch 4.5] Rune Rework General Discussion - Page 64
| Forum Index > LoL General |
|
Nos-
Canada12016 Posts
| ||
|
JazzVortical
Australia1825 Posts
| ||
|
Kyrie
1594 Posts
| ||
|
arb
Noobville17921 Posts
On April 12 2014 11:28 Sufficiency wrote: Janna's eyes are actually brown. You can zoom in from this: ![[image loading]](http://i.imgur.com/P9vsVOY.png) There are only a few pixels, but I think it's good enough to see what is going on. Like I said, the reason why that is "clearly" not Lux is the hair's lack of anti-gravity. EDIT: and the "spark" on her hat. ![[image loading]](http://i.imgur.com/mRDGv8L.png) awkward. FURTHERMORE, it could also be a different version of the Tattoo on Default Jannas arm. | ||
|
canikizu
4860 Posts
On April 12 2014 11:51 JazzVortical wrote: Syndra and Zyra number one waifus. For reasons. balls and tentacles? | ||
|
wei2coolman
United States60033 Posts
sona #1 waifu, can't talk back. | ||
|
JazzVortical
Australia1825 Posts
Syndra for her ball control. Zyra is just good looking. Plus I'm an environmentalist. | ||
|
Amui
Canada10567 Posts
On April 12 2014 11:34 Sufficiency wrote: Someone should do some math and estimate each female champion's bust size based on their splash. You know, for science. Somebody did that on reddit. | ||
|
Gahlo
United States35144 Posts
| ||
|
GolemMadness
Canada11044 Posts
| ||
|
wei2coolman
United States60033 Posts
what da fuq.CARL | ||
|
VayneAuthority
United States8983 Posts
| ||
|
Goumindong
United States3529 Posts
On April 12 2014 09:24 Sufficiency wrote: There are so many confounding factors that it'll almost always be misleading. I also believe that certain itemization choices, such as building a defensive item as a third item on ADC, is very strong. But this is very difficult to detect using the data we have available from Riot, because: 1. How do you tell it's the third item? While I can tell you what is likely to be the third item, you can't actually tell the order which a player bought the items when you only have the end-game itemization data. For example, if a Lucian has BT, TF, LW, and BV at the end of the game, you can't actually tell which item was bought third. 2. Darian's-DCAP-on-Shyvana issue, which I have previously described. When you are winning you will get more gold for more items - and your itemization choices does not really matter. 3. Issue 2 in reverse: when you are losing, you may not be able to get the item you want. Maybe you got smashed so hard in lane that you never got around to get the money for a BV, which never built first. I personally believe any meaningful itemization analysis requires the data as the game progresses, not just end-game data. I am not sure if this will ever be possible, but perhaps with in-client replay features we will be able to mine the replays for these things. W2C isn't talking about statistics. He is just talking about general qualitative analysis. He probably missed the intent of your aside when he replied to you. You're right though. There isn't any obviously good way to look at item purchases. Even if you have a measure of win probability based on the point in the match its difficult. It could be possible to use gold differential as a proxy for win probability (a pretty decent proxy for this really) and then treat the gold differential as a time series (where item purchases would be treated like structural breaks) and then the average effect of the structural break might give you some insight. But just eyeballing it this doesn't seem like it would capture the effect we want since it would only be consistent for exogenous structural breaks that had an exogenous effect. I don't know how(or if its even possible) to capture an endogenous structural break in a time series. On April 12 2014 10:58 InfSunday wrote: I thought she was Lux. Isn't Janna's hair more white? And I'm not sure if she has blue eyes either, hmm. Who is back right with purple hair? edit: Oh, it's gotta be Morg, forgot her hair was purple. Janna's tits are nearly Sona sized, so that is definitely not Janna | ||
|
Gahlo
United States35144 Posts
| ||
|
Sufficiency
Canada23833 Posts
+ Show Spoiler + http://i.imgur.com/Q2JUSb7.png The champions closer to each other have more similar "characteristics". By characteristics, I mean the following variables: - Average kills, death, assists, per minute, per game. - Average physical, magical damage to champions, per minute, per game. - Average damage taken, per minute, per game - Average Largest multikill, largest killing spree, per game. - Average Number of minions killed, number of monsters killed, per minute, per game. - Average length of CC applied, per minute, per game - What were NOT used: champion base stats, itemization choices, summoner spell choices, gold. All of these variables are then compressed onto a 2D plot which you see using PCA. The colours are done using Riot's official classification. + Show Spoiler + In case you are curious:
| ||
|
arb
Noobville17921 Posts
rofl. game says attempting to reconnect, can't connect game ends then nothing in match history | ||
|
wei2coolman
United States60033 Posts
On April 12 2014 13:32 Sufficiency wrote: I got a little bored and decided to make this picture: + Show Spoiler + http://i.imgur.com/Q2JUSb7.png The champions closer to each other have more similar "characteristics". By characteristics, I mean the following variables: - Average kills, death, assists, per minute, per game. - Average physical, magical damage to champions, per minute, per game. - Average damage taken, per minute, per game - Average Largest multikill, largest killing spree, per game. - Average Number of minions killed, number of monsters killed, per minute, per game. - Average length of CC applied, per minute, per game - What were NOT used: champion base stats, itemization choices, summoner spell choices, gold. All of these variables are then compressed onto a 2D plot which you see using PCA. The colours are done using Riot's official classification. + Show Spoiler + In case you are curious:
ADC's has closest grouping, not surprised. | ||
|
Gahlo
United States35144 Posts
On April 12 2014 13:41 arb wrote: i got drophacked in urf rofl. game says attempting to reconnect, can't connect game ends then nothing in match history Gotta protect that temporary que winrate, yo. | ||
|
Ketara
United States15065 Posts
There are so many Lux skins I would buy, but alas. None of them exist. | ||
|
Sufficiency
Canada23833 Posts
On April 12 2014 13:48 wei2coolman wrote: ADC's has closest grouping, not surprised. Yes that's sort of the plan I have in mind. I think support, mage, "tank", and fighters are grouped really well. ADCs are grouped pretty well too, and we see Corki/Kog being a bit further away to the mage's cluster due to their magic damage. The biggest issue is talon/zed/riven getting mixed up with the ADCs. It would be nice if I can separate them. | ||
| ||
