|
|
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.pngThe 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:
Comp.1 Comp.2
championsKilled 0.447
assists -0.406
numDeaths 0.212 0.143
largestKillingSpree 0.436
largestMultiKill 0.444
physicalDamageDealtToChampions 0.274 -0.383
magicDamageDealtToChampions 0.545
minionsKilled 0.353 0.199
neutralMinionsKilled -0.475
totalTimeCrowdControlDealt -0.116
damageTaken -0.503
Uh, what's the purpose of this?
|
Australia18228 Posts
On April 12 2014 08:30 Alaric wrote:
I'm not sure why people think Yi with flare is OP. It's just standard Yi for most intents and purposes.
It's just soloQ, where most people are dumb, afflicted with tunnel vision, have no coordination whatsoever and even less often the slightest grasp of drafting and how compositions work.
People have no idea how to play against Yi to begin with, it's just he kills people faster when they try to 1v1 him or refuse to hold any hard cc against him because of Flare's on-hit. But he's not really killing anyone he wouldn't kill with Wriggle's, save for maybe some duelists—who are still dumb to try and duel him anyway so it doesn't change much.
I mean, sure, he's having success. But I'm not seeing what the fuss is about, I've seen him in 5 games so far, and in each one of them it's people being normal stupid for soloQ and not Yi doing especially well. They'd lose the same way against Kha'Zix or Elise.
Also that delay definitely makes it impossible to play Orianna.
Yi in 4.4 was already quite strong, however you either kept Machete and went straight for Botrk/damage (which meant slower jungling) or invested in Wriggles which delayed your power. With Feral Flame, you get the bonus damage/gold from Wriggles while still getting the power spike since you can get Feral Flame by the time you would've originally gotten botrk.
Against a Khazix/Elise, a tanky Shyvana would be able to 1v1 either of them, but a Yi with Botrk/LW can outduel anyone (if the Yi player knows how to time his skills/items).
|
On April 12 2014 13:54 Gahlo wrote:Show nested quote +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.pngThe 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:
Comp.1 Comp.2
championsKilled 0.447
assists -0.406
numDeaths 0.212 0.143
largestKillingSpree 0.436
largestMultiKill 0.444
physicalDamageDealtToChampions 0.274 -0.383
magicDamageDealtToChampions 0.545
minionsKilled 0.353 0.199
neutralMinionsKilled -0.475
totalTimeCrowdControlDealt -0.116
damageTaken -0.503
Uh, what's the purpose of this?
The purpose is to create classifications for champions using their performance in game, instead of what Riot labels them as.
|
On April 12 2014 13:53 Ketara wrote:
This discussion has reminded me that there are no good Lux skins.
There are so many Lux skins I would buy, but alas. None of them exist.
Steel Legion Lux was OP as fuck when it first came out.
|
On April 12 2014 14:02 Sufficiency wrote:Show nested quote +On April 12 2014 13:54 Gahlo wrote: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.pngThe 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:
Comp.1 Comp.2
championsKilled 0.447
assists -0.406
numDeaths 0.212 0.143
largestKillingSpree 0.436
largestMultiKill 0.444
physicalDamageDealtToChampions 0.274 -0.383
magicDamageDealtToChampions 0.545
minionsKilled 0.353 0.199
neutralMinionsKilled -0.475
totalTimeCrowdControlDealt -0.116
damageTaken -0.503
Uh, what's the purpose of this? The purpose is to create classifications for champions using their performance in game, instead of what Riot labels them as.
I think another variable to consider is amount of damage done to turrets/buildings.
|
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
|
On April 12 2014 14:35 wei2coolman wrote:Show nested quote +On April 12 2014 14:02 Sufficiency wrote:On April 12 2014 13:54 Gahlo wrote: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.pngThe 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:
Comp.1 Comp.2
championsKilled 0.447
assists -0.406
numDeaths 0.212 0.143
largestKillingSpree 0.436
largestMultiKill 0.444
physicalDamageDealtToChampions 0.274 -0.383
magicDamageDealtToChampions 0.545
minionsKilled 0.353 0.199
neutralMinionsKilled -0.475
totalTimeCrowdControlDealt -0.116
damageTaken -0.503
Uh, what's the purpose of this? The purpose is to create classifications for champions using their performance in game, instead of what Riot labels them as. I think another variable to consider is amount of damage done to turrets/buildings.
I don't have that, unfortunately.
Only # of turrets killed.
|
On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
You can take a look at the loadings.
|
On April 12 2014 15:12 Sufficiency wrote:Show nested quote +On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
You can take a look at the loadings.
Dude that doesn't help at all lol. What are loadings?
|
does any1 here mains adc and have trouble ranking up? I was plat 1 last season and now stuck in g5 wtf, i doubt my skills have fallen and now i cant carry my team as much as i did.
|
On April 12 2014 08:30 Alaric wrote:
I'm not sure why people think Yi with flare is OP. It's just standard Yi for most intents and purposes.
It's just soloQ, where most people are dumb, afflicted with tunnel vision, have no coordination whatsoever and even less often the slightest grasp of drafting and how compositions work.
People have no idea how to play against Yi to begin with, it's just he kills people faster when they try to 1v1 him or refuse to hold any hard cc against him because of Flare's on-hit. But he's not really killing anyone he wouldn't kill with Wriggle's, save for maybe some duelists—who are still dumb to try and duel him anyway so it doesn't change much.
I mean, sure, he's having success. But I'm not seeing what the fuss is about, I've seen him in 5 games so far, and in each one of them it's people being normal stupid for soloQ and not Yi doing especially well. They'd lose the same way against Kha'Zix or Elise.
Also that delay definitely makes it impossible to play Orianna.
That's not true ... if you even tried the new Yi build into Youmuu's, it is noticeably stronger than Yi's jungling has been before. I practically thought there was a bug with FF due to how much it was doing. ADC's/AP's were being erased in two hits from full.
|
On April 12 2014 15:25 justiceknight wrote:
does any1 here mains adc and have trouble ranking up? I was plat 1 last season and now stuck in g5 wtf, i doubt my skills have fallen and now i cant carry my team as much as i did.
Welcome to S4 ADC.
|
On April 12 2014 15:19 JazzVortical wrote:Show nested quote +On April 12 2014 15:12 Sufficiency wrote:On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
You can take a look at the loadings. Dude that doesn't help at all lol. What are loadings?
Comp.1 Comp.2
championsKilled 0.447
assists -0.406
numDeaths 0.212 0.143
largestKillingSpree 0.436
largestMultiKill 0.444
physicalDamageDealtToChampions 0.274 -0.383
magicDamageDealtToChampions 0.545
minionsKilled 0.353 0.199
neutralMinionsKilled -0.475
totalTimeCrowdControlDealt -0.116
damageTaken -0.503
|
On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
Prolly they have a unique mechanic.
|
On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
Probably things such as low minion damage, low neutral minion damage.
but very high multikill counts, tons of magic damage ton to champion. low physical damage.
oddly skewed K/D/A counts due to nature of resets, and what happens when kat falls behind.
she's just got a pretty unique kit in how it plays out compared to the rest of the other champions.
interesting enough, the outliers seem to be assassins.
fizz/kat/akali/fiora/master yi/k6.
more interesting variables that could be added, is first dragon timing, first baron timing. (speaks a lot about power curve).
|
I really feel like Yi is a time bomb like Ryze or something in that its only a matter of time before he gets enough farm/items to shit on everyone single handedly tbh.
|
On April 12 2014 15:42 Gahlo wrote:Show nested quote +On April 12 2014 15:25 justiceknight wrote:
does any1 here mains adc and have trouble ranking up? I was plat 1 last season and now stuck in g5 wtf, i doubt my skills have fallen and now i cant carry my team as much as i did. Welcome to S4 ADC.
4 of the top 5 in challenger are ADC's.
|
On April 12 2014 15:52 Sufficiency wrote:Show nested quote +On April 12 2014 15:19 JazzVortical wrote:On April 12 2014 15:12 Sufficiency wrote:On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
You can take a look at the loadings. Dude that doesn't help at all lol. What are loadings?
Comp.1 Comp.2
championsKilled 0.447
assists -0.406
numDeaths 0.212 0.143
largestKillingSpree 0.436
largestMultiKill 0.444
physicalDamageDealtToChampions 0.274 -0.383
magicDamageDealtToChampions 0.545
minionsKilled 0.353 0.199
neutralMinionsKilled -0.475
totalTimeCrowdControlDealt -0.116
damageTaken -0.503
Im sorry mate, maybe I wasn't clear. This is just numbers and doesn't help me at all. I wanted what we learn from the graph.
What's the explanation behind Katarina being on her own? What information can we glean? My thanks to clutz and Wei  they got what I meant.
|
On April 12 2014 15:54 wei2coolman wrote:Show nested quote +On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
Probably things such as low minion damage, low neutral minion damage. but very high multikill counts, tons of magic damage ton to champion. low physical damage. oddly skewed K/D/A counts due to nature of resets, and what happens when kat falls behind. she's just got a pretty unique kit in how it plays out compared to the rest of the other champions. interesting enough, the outliers seem to be assassins. fizz/kat/akali/fiora/master yi/k6. more interesting variables that could be added, is first dragon timing, first baron timing. (speaks a lot about power curve).
Yeah.
The horizontal axis is for high kills, high killing streak, low assist, high physical damage. This is why Yi has like the highest value here.
The vertical axis is for high magic damage and low physical damage.
Katarina is in the unique situation to be in both categories. High kills, low assist, AND magic damage.
Unfortunately, the method I used (PCA) gives results that are hard to interpret. This is a known flaw of this method.
EDIT: no data on first dragon, etc 
On April 12 2014 16:34 JazzVortical wrote:Show nested quote +On April 12 2014 15:52 Sufficiency wrote:On April 12 2014 15:19 JazzVortical wrote:On April 12 2014 15:12 Sufficiency wrote:On April 12 2014 14:58 JazzVortical wrote:
So, this might be dumb but I'm terrible with stats and maths and what not:
What inferences do we get about champs like Katarina, who sits way out on her own?
You can take a look at the loadings. Dude that doesn't help at all lol. What are loadings?
Comp.1 Comp.2
championsKilled 0.447
assists -0.406
numDeaths 0.212 0.143
largestKillingSpree 0.436
largestMultiKill 0.444
physicalDamageDealtToChampions 0.274 -0.383
magicDamageDealtToChampions 0.545
minionsKilled 0.353 0.199
neutralMinionsKilled -0.475
totalTimeCrowdControlDealt -0.116
damageTaken -0.503
Im sorry mate, maybe I wasn't clear. This is just numbers and doesn't help me at all. I wanted what we learn from the graph. What's the explanation behind Katarina being on her own? What information can we glean? My thanks to clutz and Wei they got what I meant.
This may sound absurd, but you can't actually learn too much, beyond each champion's relative position.
Originally, I had this table of data with 118 rows (one for each champion), and 11 columns (one for each of the variables above). What I did, essentially, is that I condensed 11 column worth of data into 2 columns. This allows me to easily plot the data onto that picture which you saw. The reason why someone may want to do this is to be able to visualize the data (think about how you would plot 11 columns of data). By looking at the plot, we see that we can sort of cluster the champions into their categories (mage, tank, support, fighter, etc.) by using a single plot from these 11 columns of data.
The disadvantage of this method is that when you condense 11 columns into 2, you are just taking chunks of each column and piecing them together (a linear combination). This makes interpretation difficult.
There are other methods which, for example, allows you to condense your data but each new column can only be made from a limited number of columns of the original data. But those are usually for data that has many, many columns.
|
On April 12 2014 13:53 Ketara wrote:
This discussion has reminded me that there are no good Lux skins.
There are so many Lux skins I would buy, but alas. None of them exist.
You've already displayed your lack of taste by stating over and over that you like Lux. Why would you even be allowed to talk about skins?
|
|
|
|
|
|
EPT
