|
|
|
On March 27 2014 23:33 JimmiC wrote:Show nested quote +On March 27 2014 08:49 IMoperator wrote:
who wins a level 3 duel, panth or lee sin? Voli does with his passive up. Especially in lane because you can E and fight in minions which do a sick amount of damage early and his won't hit you whiles your hit him. edit: I get a F for reading comprehension, I read it as who bats both in duels =-(
Should be lee sin, theres very few jungles who can stand toe to toe with him in early duels like that
|
United Kingdom50293 Posts
On March 27 2014 23:39 JimmiC wrote:Show nested quote +On March 27 2014 12:31 onlywonderboy wrote:
Scarra said he recognized my handle from posting our stuff on Reddit when we met him at Worlds. And he's actually retweeted NA content so he's pretty much my favorite (former)pro player. Pretty amazing how that works eh? Al the people in this forum that are trying to make it pro or make a living streaming keep that in mind. To make a fan for life all you need to do is interact with the person. Makes us feel special and we will support you for ever. In sc2 its how guys like destiny and MaximusBlack get good numbers even though they were never pro skill level. Tons of interaction.
Excuse me good sir, destiny 2-1 bomber never forget.
|
On March 28 2014 00:55 Fusilero wrote:Show nested quote +On March 27 2014 23:39 JimmiC wrote:On March 27 2014 12:31 onlywonderboy wrote:
Scarra said he recognized my handle from posting our stuff on Reddit when we met him at Worlds. And he's actually retweeted NA content so he's pretty much my favorite (former)pro player. Pretty amazing how that works eh? Al the people in this forum that are trying to make it pro or make a living streaming keep that in mind. To make a fan for life all you need to do is interact with the person. Makes us feel special and we will support you for ever. In sc2 its how guys like destiny and MaximusBlack get good numbers even though they were never pro skill level. Tons of interaction. Excuse me good sir, destiny 2-1 bomber never forget.
he was still terrible and raged too much
|
On March 27 2014 14:32 nobodywonder wrote:
do you guys yolo/ go full retard in norms usually? like i only do stuff like adc fiddle, blue build jax, afk split push cait, and forever feeding singed
it's kinda pissing off my lower elo friends who are outraged at my feeding and huehuehue
You expected me to carry you?
Apologies my good sir, I thought this was normals.
|
On March 27 2014 23:24 Sufficiency wrote:Show nested quote +On March 27 2014 21:38 Fwmeh wrote:On March 27 2014 02:52 Sufficiency wrote:On March 27 2014 02:47 Eiii wrote:But wouldn't you rather see... average number of pentakills per normal game?!? + Show Spoiler ++---------------+--------+
| champion_name | num |
+---------------+--------+
| MasterYi | 0.0111 |
| Katarina | 0.0109 |
| Akali | 0.0064 |
| Jinx | 0.0051 |
| Draven | 0.0050 |
| Yasuo | 0.0049 |
| Fiora | 0.0048 |
| Vayne | 0.0046 |
| Tristana | 0.0039 |
| Khazix | 0.0038 |
| Mordekaiser | 0.0034 |
| Twitch | 0.0033 |
| Corki | 0.0030 |
| Lucian | 0.0029 |
| Quinn | 0.0028 |
| Graves | 0.0028 |
| Varus | 0.0025 |
| Jayce | 0.0024 |
| Riven | 0.0024 |
| Ezreal | 0.0023 |
| Cassiopeia | 0.0022 |
| Caitlyn | 0.0022 |
| Talon | 0.0021 |
| Sivir | 0.0020 |
| Swain | 0.0020 |
| MissFortune | 0.0019 |
| Ashe | 0.0018 |
| KogMaw | 0.0017 |
| Kassadin | 0.0017 |
| Tryndamere | 0.0017 |
| Fizz | 0.0017 |
| Jax | 0.0016 |
| Kayle | 0.0016 |
| Irelia | 0.0016 |
| Diana | 0.0015 |
| Darius | 0.0015 |
| Rengar | 0.0014 |
| Sion | 0.0014 |
| Zed | 0.0013 |
| Rumble | 0.0011 |
| FiddleSticks | 0.0011 |
| Viktor | 0.0011 |
| Teemo | 0.0011 |
| Ryze | 0.0011 |
| Pantheon | 0.0011 |
| Ziggs | 0.0010 |
| Gangplank | 0.0010 |
| Aatrox | 0.0010 |
| TwistedFate | 0.0009 |
| Evelynn | 0.0009 |
| Malzahar | 0.0009 |
| Orianna | 0.0009 |
| Poppy | 0.0008 |
| MonkeyKing | 0.0008 |
| Ahri | 0.0008 |
| Nocturne | 0.0008 |
| Vladimir | 0.0008 |
| Leblanc | 0.0008 |
| Anivia | 0.0008 |
| Nasus | 0.0008 |
| Gragas | 0.0007 |
| Brand | 0.0007 |
| XinZhao | 0.0007 |
| LeeSin | 0.0007 |
| Kennen | 0.0007 |
| Vi | 0.0006 |
| Xerath | 0.0006 |
| Skarner | 0.0006 |
| Syndra | 0.0006 |
| Karthus | 0.0006 |
| Veigar | 0.0006 |
| Volibear | 0.0006 |
| Shyvana | 0.0005 |
| Heimerdinger | 0.0005 |
| Shaco | 0.0005 |
| Olaf | 0.0005 |
| Annie | 0.0005 |
| Warwick | 0.0005 |
| Yorick | 0.0004 |
| Lux | 0.0004 |
| Amumu | 0.0004 |
| Nidalee | 0.0004 |
| Nunu | 0.0004 |
| Lulu | 0.0003 |
| Lissandra | 0.0003 |
| Renekton | 0.0003 |
| Hecarim | 0.0003 |
| Udyr | 0.0003 |
| Singed | 0.0002 |
| Rammus | 0.0002 |
| Elise | 0.0002 |
| JarvanIV | 0.0002 |
| Malphite | 0.0002 |
| Trundle | 0.0002 |
| Morgana | 0.0001 |
| DrMundo | 0.0001 |
| Garen | 0.0001 |
| Shen | 0.0001 |
| Thresh | 0.0001 |
| Alistar | 0.0001 |
| Nami | 0.0000 |
| Blitzcrank | 0.0000 |
| Galio | 0.0000 |
| Taric | 0.0000 |
| Sejuani | 0.0000 |
| Zyra | 0.0000 |
| Urgot | 0.0000 |
| Nautilus | 0.0000 |
| Sona | 0.0000 |
| Soraka | 0.0000 |
| Chogath | 0.0000 |
| Leona | 0.0000 |
| Karma | 0.0000 |
| Zac | 0.0000 |
| Janna | 0.0000 |
| Maokai | 0.0000 |
| Zilean | 0.0000 |
+---------------+--------+ By the way sufficiency, want to join forces on this whole data collection thing? Sure, but I didn't use databases... I am just dumping the data into CSV files... But yeah, I was thinking about the average number of pentas per champion, which weights the popularity of the champion. I was too lazy to do it though... Are you doing/planning to do some association rule mining on the data? What formating do you use? Association rule will be interesting, but that is not what I am doing. There are a few things that can be done. One of them is champion counter / synergy analysis. I am using logistic regression for that.
I just thought that rule mining would be an easy way to find counters/synergies, if the data could feasibly be transformed to suit that. Don't have much experience with logistic regression myself.
|
|
|
On March 28 2014 01:12 Fwmeh wrote:Show nested quote +On March 27 2014 23:24 Sufficiency wrote:On March 27 2014 21:38 Fwmeh wrote:On March 27 2014 02:52 Sufficiency wrote:On March 27 2014 02:47 Eiii wrote:But wouldn't you rather see... average number of pentakills per normal game?!? + Show Spoiler ++---------------+--------+
| champion_name | num |
+---------------+--------+
| MasterYi | 0.0111 |
| Katarina | 0.0109 |
| Akali | 0.0064 |
| Jinx | 0.0051 |
| Draven | 0.0050 |
| Yasuo | 0.0049 |
| Fiora | 0.0048 |
| Vayne | 0.0046 |
| Tristana | 0.0039 |
| Khazix | 0.0038 |
| Mordekaiser | 0.0034 |
| Twitch | 0.0033 |
| Corki | 0.0030 |
| Lucian | 0.0029 |
| Quinn | 0.0028 |
| Graves | 0.0028 |
| Varus | 0.0025 |
| Jayce | 0.0024 |
| Riven | 0.0024 |
| Ezreal | 0.0023 |
| Cassiopeia | 0.0022 |
| Caitlyn | 0.0022 |
| Talon | 0.0021 |
| Sivir | 0.0020 |
| Swain | 0.0020 |
| MissFortune | 0.0019 |
| Ashe | 0.0018 |
| KogMaw | 0.0017 |
| Kassadin | 0.0017 |
| Tryndamere | 0.0017 |
| Fizz | 0.0017 |
| Jax | 0.0016 |
| Kayle | 0.0016 |
| Irelia | 0.0016 |
| Diana | 0.0015 |
| Darius | 0.0015 |
| Rengar | 0.0014 |
| Sion | 0.0014 |
| Zed | 0.0013 |
| Rumble | 0.0011 |
| FiddleSticks | 0.0011 |
| Viktor | 0.0011 |
| Teemo | 0.0011 |
| Ryze | 0.0011 |
| Pantheon | 0.0011 |
| Ziggs | 0.0010 |
| Gangplank | 0.0010 |
| Aatrox | 0.0010 |
| TwistedFate | 0.0009 |
| Evelynn | 0.0009 |
| Malzahar | 0.0009 |
| Orianna | 0.0009 |
| Poppy | 0.0008 |
| MonkeyKing | 0.0008 |
| Ahri | 0.0008 |
| Nocturne | 0.0008 |
| Vladimir | 0.0008 |
| Leblanc | 0.0008 |
| Anivia | 0.0008 |
| Nasus | 0.0008 |
| Gragas | 0.0007 |
| Brand | 0.0007 |
| XinZhao | 0.0007 |
| LeeSin | 0.0007 |
| Kennen | 0.0007 |
| Vi | 0.0006 |
| Xerath | 0.0006 |
| Skarner | 0.0006 |
| Syndra | 0.0006 |
| Karthus | 0.0006 |
| Veigar | 0.0006 |
| Volibear | 0.0006 |
| Shyvana | 0.0005 |
| Heimerdinger | 0.0005 |
| Shaco | 0.0005 |
| Olaf | 0.0005 |
| Annie | 0.0005 |
| Warwick | 0.0005 |
| Yorick | 0.0004 |
| Lux | 0.0004 |
| Amumu | 0.0004 |
| Nidalee | 0.0004 |
| Nunu | 0.0004 |
| Lulu | 0.0003 |
| Lissandra | 0.0003 |
| Renekton | 0.0003 |
| Hecarim | 0.0003 |
| Udyr | 0.0003 |
| Singed | 0.0002 |
| Rammus | 0.0002 |
| Elise | 0.0002 |
| JarvanIV | 0.0002 |
| Malphite | 0.0002 |
| Trundle | 0.0002 |
| Morgana | 0.0001 |
| DrMundo | 0.0001 |
| Garen | 0.0001 |
| Shen | 0.0001 |
| Thresh | 0.0001 |
| Alistar | 0.0001 |
| Nami | 0.0000 |
| Blitzcrank | 0.0000 |
| Galio | 0.0000 |
| Taric | 0.0000 |
| Sejuani | 0.0000 |
| Zyra | 0.0000 |
| Urgot | 0.0000 |
| Nautilus | 0.0000 |
| Sona | 0.0000 |
| Soraka | 0.0000 |
| Chogath | 0.0000 |
| Leona | 0.0000 |
| Karma | 0.0000 |
| Zac | 0.0000 |
| Janna | 0.0000 |
| Maokai | 0.0000 |
| Zilean | 0.0000 |
+---------------+--------+ By the way sufficiency, want to join forces on this whole data collection thing? Sure, but I didn't use databases... I am just dumping the data into CSV files... But yeah, I was thinking about the average number of pentas per champion, which weights the popularity of the champion. I was too lazy to do it though... Are you doing/planning to do some association rule mining on the data? What formating do you use? Association rule will be interesting, but that is not what I am doing. There are a few things that can be done. One of them is champion counter / synergy analysis. I am using logistic regression for that. I just thought that rule mining would be an easy way to find counters/synergies, if the data could feasibly be transformed to suit that. Don't have much experience with logistic regression myself.
Association rule itself is fine, but you have to account for the overall power of a champion.
For example, when you take a champion that wins around 50% of the time overall (say, Ashe), and pit her against a "bad" champion such as Twisted Fate, you will see significant increase in Ashe's win rate - simply because Twisted Fate is bad, not because Ashe counters Twisted Fate in particular. Adjustments have to be made to account for the overall power of the two champions in question.
|
I woke up a few hours before qtpie started streaming yesterday, and he's already streaming today and I haven't slept...
|
United States47024 Posts
It's also just more complex than that. Champion A might be a common pick in teamcomps that are good against champion B, but that doesn't mean that picking champion A in a vacuum will be good against champion B.
It's just muddled in general because you're basically using the working assumption that a team's strength is the sum of its components, when that's not remotely close to being true in this game.
|
On March 28 2014 01:29 Sufficiency wrote:Show nested quote +On March 28 2014 01:12 Fwmeh wrote:On March 27 2014 23:24 Sufficiency wrote:On March 27 2014 21:38 Fwmeh wrote:On March 27 2014 02:52 Sufficiency wrote:On March 27 2014 02:47 Eiii wrote:But wouldn't you rather see... average number of pentakills per normal game?!? + Show Spoiler ++---------------+--------+
| champion_name | num |
+---------------+--------+
| MasterYi | 0.0111 |
| Katarina | 0.0109 |
| Akali | 0.0064 |
| Jinx | 0.0051 |
| Draven | 0.0050 |
| Yasuo | 0.0049 |
| Fiora | 0.0048 |
| Vayne | 0.0046 |
| Tristana | 0.0039 |
| Khazix | 0.0038 |
| Mordekaiser | 0.0034 |
| Twitch | 0.0033 |
| Corki | 0.0030 |
| Lucian | 0.0029 |
| Quinn | 0.0028 |
| Graves | 0.0028 |
| Varus | 0.0025 |
| Jayce | 0.0024 |
| Riven | 0.0024 |
| Ezreal | 0.0023 |
| Cassiopeia | 0.0022 |
| Caitlyn | 0.0022 |
| Talon | 0.0021 |
| Sivir | 0.0020 |
| Swain | 0.0020 |
| MissFortune | 0.0019 |
| Ashe | 0.0018 |
| KogMaw | 0.0017 |
| Kassadin | 0.0017 |
| Tryndamere | 0.0017 |
| Fizz | 0.0017 |
| Jax | 0.0016 |
| Kayle | 0.0016 |
| Irelia | 0.0016 |
| Diana | 0.0015 |
| Darius | 0.0015 |
| Rengar | 0.0014 |
| Sion | 0.0014 |
| Zed | 0.0013 |
| Rumble | 0.0011 |
| FiddleSticks | 0.0011 |
| Viktor | 0.0011 |
| Teemo | 0.0011 |
| Ryze | 0.0011 |
| Pantheon | 0.0011 |
| Ziggs | 0.0010 |
| Gangplank | 0.0010 |
| Aatrox | 0.0010 |
| TwistedFate | 0.0009 |
| Evelynn | 0.0009 |
| Malzahar | 0.0009 |
| Orianna | 0.0009 |
| Poppy | 0.0008 |
| MonkeyKing | 0.0008 |
| Ahri | 0.0008 |
| Nocturne | 0.0008 |
| Vladimir | 0.0008 |
| Leblanc | 0.0008 |
| Anivia | 0.0008 |
| Nasus | 0.0008 |
| Gragas | 0.0007 |
| Brand | 0.0007 |
| XinZhao | 0.0007 |
| LeeSin | 0.0007 |
| Kennen | 0.0007 |
| Vi | 0.0006 |
| Xerath | 0.0006 |
| Skarner | 0.0006 |
| Syndra | 0.0006 |
| Karthus | 0.0006 |
| Veigar | 0.0006 |
| Volibear | 0.0006 |
| Shyvana | 0.0005 |
| Heimerdinger | 0.0005 |
| Shaco | 0.0005 |
| Olaf | 0.0005 |
| Annie | 0.0005 |
| Warwick | 0.0005 |
| Yorick | 0.0004 |
| Lux | 0.0004 |
| Amumu | 0.0004 |
| Nidalee | 0.0004 |
| Nunu | 0.0004 |
| Lulu | 0.0003 |
| Lissandra | 0.0003 |
| Renekton | 0.0003 |
| Hecarim | 0.0003 |
| Udyr | 0.0003 |
| Singed | 0.0002 |
| Rammus | 0.0002 |
| Elise | 0.0002 |
| JarvanIV | 0.0002 |
| Malphite | 0.0002 |
| Trundle | 0.0002 |
| Morgana | 0.0001 |
| DrMundo | 0.0001 |
| Garen | 0.0001 |
| Shen | 0.0001 |
| Thresh | 0.0001 |
| Alistar | 0.0001 |
| Nami | 0.0000 |
| Blitzcrank | 0.0000 |
| Galio | 0.0000 |
| Taric | 0.0000 |
| Sejuani | 0.0000 |
| Zyra | 0.0000 |
| Urgot | 0.0000 |
| Nautilus | 0.0000 |
| Sona | 0.0000 |
| Soraka | 0.0000 |
| Chogath | 0.0000 |
| Leona | 0.0000 |
| Karma | 0.0000 |
| Zac | 0.0000 |
| Janna | 0.0000 |
| Maokai | 0.0000 |
| Zilean | 0.0000 |
+---------------+--------+ By the way sufficiency, want to join forces on this whole data collection thing? Sure, but I didn't use databases... I am just dumping the data into CSV files... But yeah, I was thinking about the average number of pentas per champion, which weights the popularity of the champion. I was too lazy to do it though... Are you doing/planning to do some association rule mining on the data? What formating do you use? Association rule will be interesting, but that is not what I am doing. There are a few things that can be done. One of them is champion counter / synergy analysis. I am using logistic regression for that. I just thought that rule mining would be an easy way to find counters/synergies, if the data could feasibly be transformed to suit that. Don't have much experience with logistic regression myself. Association rule itself is fine, but you have to account for the overall power of a champion. For example, when you take a champion that wins around 50% of the time overall (say, Ashe), and pit her against a "bad" champion such as Twisted Fate, you will see significant increase in Ashe's win rate - simply because Twisted Fate is bad, not because Ashe counters Twisted Fate in particular. Adjustments have to be made to account for the overall power of the two champions in question.
I don't think TF is bad in general, but in this meta i feel like he really suffers. Played him the other day and still feel like he's good at what he was before. I.e porting in and shitting on a single targets life or stopping tower sieges
|
On March 28 2014 01:47 TheYango wrote:
It's also just more complex than that. Champion A might be a common pick in teamcomps that are good against champion B, but that doesn't mean that picking champion A in a vacuum will be good against champion B.
It's just muddled in general because you're basically using the working assumption that a team's strength is the sum of its components, when that's not remotely close to being true in this game.
Of course, but reasonably simplifying the model allows easier analysis of the data that requires less work.
|
Assuming that comps are designed somewhat well, isn't the principle of synergy that they're enhancing each others' strengths? For example, Lulu could pair well with Nidalee/Ziggs because she provides peel, poke, disengage and some kind of counter-initiation with her ult. So Lulu could be common in poke comps, but on the other hand she's good by herself against champions vulnerable to poke (like Udyr or Nasus for example).
Sivir and Karma make a lot of "facerush" champions stronger, but champions vulnerable to "zerg comps" like Jinx or Kog'Maw should already be somewhat vulnerable to divers like Shyvana or Aatrox, shouldn't they?
Of course there's the point that soloQ is far from being synergy-savvy, cf. the amount of Leona/Vi+Nidalee combo that pop up.
I'd argue that the composition of teams apart from the two champions would need to get taken into account too, but I assume that the model already relies on having large sample sizes to consider that for example in the Ashe vs Yasuo match-up, Amumu pops up as often in Ashe's team than in Yasuo's.
|
On March 28 2014 02:07 Alaric wrote:
Assuming that comps are designed somewhat well, isn't the principle of synergy that they're enhancing each others' strengths? For example, Lulu could pair well with Nidalee/Ziggs because she provides peel, poke, disengage and some kind of counter-initiation with her ult. So Lulu could be common in poke comps, but on the other hand she's good by herself against champions vulnerable to poke (like Udyr or Nasus for example).
Sivir and Karma make a lot of "facerush" champions stronger, but champions vulnerable to "zerg comps" like Jinx or Kog'Maw should already be somewhat vulnerable to divers like Shyvana or Aatrox, shouldn't they?
Of course there's the point that soloQ is far from being synergy-savvy, cf. the amount of Leona/Vi+Nidalee combo that pop up.
I'd argue that the composition of teams apart from the two champions would need to get taken into account too, but I assume that the model already relies on having large sample sizes to consider that for example in the Ashe vs Yasuo match-up, Amumu pops up as often in Ashe's team than in Yasuo's.
I feel like outside of a poke comp Nidalee literally brings nothing to her team at any stage of the game personally
|
On March 28 2014 02:11 arb wrote:Show nested quote +On March 28 2014 02:07 Alaric wrote:
Assuming that comps are designed somewhat well, isn't the principle of synergy that they're enhancing each others' strengths? For example, Lulu could pair well with Nidalee/Ziggs because she provides peel, poke, disengage and some kind of counter-initiation with her ult. So Lulu could be common in poke comps, but on the other hand she's good by herself against champions vulnerable to poke (like Udyr or Nasus for example).
Sivir and Karma make a lot of "facerush" champions stronger, but champions vulnerable to "zerg comps" like Jinx or Kog'Maw should already be somewhat vulnerable to divers like Shyvana or Aatrox, shouldn't they?
Of course there's the point that soloQ is far from being synergy-savvy, cf. the amount of Leona/Vi+Nidalee combo that pop up.
I'd argue that the composition of teams apart from the two champions would need to get taken into account too, but I assume that the model already relies on having large sample sizes to consider that for example in the Ashe vs Yasuo match-up, Amumu pops up as often in Ashe's team than in Yasuo's. I feel like outside of a poke comp Nidalee literally brings nothing to her team at any stage of the game personally
Nidalee is always terror in soloq unless you just happen to be a coordinated team for some reason.Also fuck her headhunter skin the spears are pretty much invisible.
|
On March 28 2014 02:11 arb wrote:Show nested quote +On March 28 2014 02:07 Alaric wrote:
Assuming that comps are designed somewhat well, isn't the principle of synergy that they're enhancing each others' strengths? For example, Lulu could pair well with Nidalee/Ziggs because she provides peel, poke, disengage and some kind of counter-initiation with her ult. So Lulu could be common in poke comps, but on the other hand she's good by herself against champions vulnerable to poke (like Udyr or Nasus for example).
Sivir and Karma make a lot of "facerush" champions stronger, but champions vulnerable to "zerg comps" like Jinx or Kog'Maw should already be somewhat vulnerable to divers like Shyvana or Aatrox, shouldn't they?
Of course there's the point that soloQ is far from being synergy-savvy, cf. the amount of Leona/Vi+Nidalee combo that pop up.
I'd argue that the composition of teams apart from the two champions would need to get taken into account too, but I assume that the model already relies on having large sample sizes to consider that for example in the Ashe vs Yasuo match-up, Amumu pops up as often in Ashe's team than in Yasuo's. I feel like outside of a poke comp Nidalee literally brings nothing to her team at any stage of the game personally
She can split push ok, though not as good as others.
|
Her heal is pretty strong in push comps thanks to the AS buff, and the high ratio also helps her protect her team when against assassins. In the super-tanky top lane meta, having a 40% resistances shred that lasts 8 seconds is arguably stronger than Trundle's ult if you manage to use them well, and they also give some kind of map control through vision.
Of course, the only relevant point is that she turns any comp with wave clear into a poke comp by herself though.
|
On March 28 2014 01:47 TheYango wrote:
It's also just more complex than that. Champion A might be a common pick in teamcomps that are good against champion B, but that doesn't mean that picking champion A in a vacuum will be good against champion B.
It's just muddled in general because you're basically using the working assumption that a team's strength is the sum of its components, when that's not remotely close to being true in this game.
Gives me an interesting idea (which would be hard to do because of the number of champs). Five man lineups. They are some of the best stats in basketball.
|
On March 28 2014 01:47 TheYango wrote:
It's also just more complex than that. Champion A might be a common pick in teamcomps that are good against champion B, but that doesn't mean that picking champion A in a vacuum will be good against champion B.
It's just muddled in general because you're basically using the working assumption that a team's strength is the sum of its components, when that's not remotely close to being true in this game.
The problem is that the number of possible compositions and interaction terms will actually eat our degrees of freedom and make the calculation take way too long.
In this case, assuming away those issue, even if you don't believe it, is probably the best option. So long as you state the caveat
|
On March 28 2014 01:29 Sufficiency wrote:Show nested quote +On March 28 2014 01:12 Fwmeh wrote:On March 27 2014 23:24 Sufficiency wrote:On March 27 2014 21:38 Fwmeh wrote:On March 27 2014 02:52 Sufficiency wrote:On March 27 2014 02:47 Eiii wrote:But wouldn't you rather see... average number of pentakills per normal game?!? + Show Spoiler ++---------------+--------+
| champion_name | num |
+---------------+--------+
| MasterYi | 0.0111 |
| Katarina | 0.0109 |
| Akali | 0.0064 |
| Jinx | 0.0051 |
| Draven | 0.0050 |
| Yasuo | 0.0049 |
| Fiora | 0.0048 |
| Vayne | 0.0046 |
| Tristana | 0.0039 |
| Khazix | 0.0038 |
| Mordekaiser | 0.0034 |
| Twitch | 0.0033 |
| Corki | 0.0030 |
| Lucian | 0.0029 |
| Quinn | 0.0028 |
| Graves | 0.0028 |
| Varus | 0.0025 |
| Jayce | 0.0024 |
| Riven | 0.0024 |
| Ezreal | 0.0023 |
| Cassiopeia | 0.0022 |
| Caitlyn | 0.0022 |
| Talon | 0.0021 |
| Sivir | 0.0020 |
| Swain | 0.0020 |
| MissFortune | 0.0019 |
| Ashe | 0.0018 |
| KogMaw | 0.0017 |
| Kassadin | 0.0017 |
| Tryndamere | 0.0017 |
| Fizz | 0.0017 |
| Jax | 0.0016 |
| Kayle | 0.0016 |
| Irelia | 0.0016 |
| Diana | 0.0015 |
| Darius | 0.0015 |
| Rengar | 0.0014 |
| Sion | 0.0014 |
| Zed | 0.0013 |
| Rumble | 0.0011 |
| FiddleSticks | 0.0011 |
| Viktor | 0.0011 |
| Teemo | 0.0011 |
| Ryze | 0.0011 |
| Pantheon | 0.0011 |
| Ziggs | 0.0010 |
| Gangplank | 0.0010 |
| Aatrox | 0.0010 |
| TwistedFate | 0.0009 |
| Evelynn | 0.0009 |
| Malzahar | 0.0009 |
| Orianna | 0.0009 |
| Poppy | 0.0008 |
| MonkeyKing | 0.0008 |
| Ahri | 0.0008 |
| Nocturne | 0.0008 |
| Vladimir | 0.0008 |
| Leblanc | 0.0008 |
| Anivia | 0.0008 |
| Nasus | 0.0008 |
| Gragas | 0.0007 |
| Brand | 0.0007 |
| XinZhao | 0.0007 |
| LeeSin | 0.0007 |
| Kennen | 0.0007 |
| Vi | 0.0006 |
| Xerath | 0.0006 |
| Skarner | 0.0006 |
| Syndra | 0.0006 |
| Karthus | 0.0006 |
| Veigar | 0.0006 |
| Volibear | 0.0006 |
| Shyvana | 0.0005 |
| Heimerdinger | 0.0005 |
| Shaco | 0.0005 |
| Olaf | 0.0005 |
| Annie | 0.0005 |
| Warwick | 0.0005 |
| Yorick | 0.0004 |
| Lux | 0.0004 |
| Amumu | 0.0004 |
| Nidalee | 0.0004 |
| Nunu | 0.0004 |
| Lulu | 0.0003 |
| Lissandra | 0.0003 |
| Renekton | 0.0003 |
| Hecarim | 0.0003 |
| Udyr | 0.0003 |
| Singed | 0.0002 |
| Rammus | 0.0002 |
| Elise | 0.0002 |
| JarvanIV | 0.0002 |
| Malphite | 0.0002 |
| Trundle | 0.0002 |
| Morgana | 0.0001 |
| DrMundo | 0.0001 |
| Garen | 0.0001 |
| Shen | 0.0001 |
| Thresh | 0.0001 |
| Alistar | 0.0001 |
| Nami | 0.0000 |
| Blitzcrank | 0.0000 |
| Galio | 0.0000 |
| Taric | 0.0000 |
| Sejuani | 0.0000 |
| Zyra | 0.0000 |
| Urgot | 0.0000 |
| Nautilus | 0.0000 |
| Sona | 0.0000 |
| Soraka | 0.0000 |
| Chogath | 0.0000 |
| Leona | 0.0000 |
| Karma | 0.0000 |
| Zac | 0.0000 |
| Janna | 0.0000 |
| Maokai | 0.0000 |
| Zilean | 0.0000 |
+---------------+--------+ By the way sufficiency, want to join forces on this whole data collection thing? Sure, but I didn't use databases... I am just dumping the data into CSV files... But yeah, I was thinking about the average number of pentas per champion, which weights the popularity of the champion. I was too lazy to do it though... Are you doing/planning to do some association rule mining on the data? What formating do you use? Association rule will be interesting, but that is not what I am doing. There are a few things that can be done. One of them is champion counter / synergy analysis. I am using logistic regression for that. I just thought that rule mining would be an easy way to find counters/synergies, if the data could feasibly be transformed to suit that. Don't have much experience with logistic regression myself. Association rule itself is fine, but you have to account for the overall power of a champion. For example, when you take a champion that wins around 50% of the time overall (say, Ashe), and pit her against a "bad" champion such as Twisted Fate, you will see significant increase in Ashe's win rate - simply because Twisted Fate is bad, not because Ashe counters Twisted Fate in particular. Adjustments have to be made to account for the overall power of the two champions in question.
With regards to counters, it would be reasonable to look for rules indicating that a champion loses more games against one particular champion, compared to some baseline. Then you just use whatever minconf you feels is enough. I think it would be pretty straight forward, though I have not spent much time on it...
|
|
|
|
|
|
EPT
