| Forum Index > SC2 General |
|
xian_
Germany12 Posts
Hope you can gather enough data to calculate solid numbers for master+ soon. In the end that is the level most people here care about. I wonder wheter we will be able to see changes caused by future patches as well. Or even see which regions can adopt faster to changes. Keep it up. ps: If you expand your OP with information regarding the methods you might be able to avoid some discussions. | ||
|
lolcanoe
United States57 Posts
On July 11 2012 09:34 lolcanoe wrote: I'm confused here - why are there 4 random groups? Why not just lump them together? But im still to 99,31 % secure Terran is underpowered -30.09 Where the hell are you getting these numbers for? I'm assuming this 99.3 comes from some application of the normal distribution - have you been thorough established normality in MMR? Based on my understanding of the MMR system, you would not expect normality... As a stats junky, I'm very suspicious of the statistics here. 4 because i had 4 races ( with random ) 2 run had only 3 . Dont have be this number only to get kind of close to the amount of accounts of the races. MMR is normal by definition ( if nothing goes wrong) If this data is normal i can not tell i did not test yet Its just the % of the chance that random data would create such an result. depending on tests. Im not 100% sure if you can do it this way but i dont see why not. Im no statistic guru in any way. I am not terrible in it but its many years ago i worked with it. Basic i just produce random data to show that the data is not biased and how "stable" it is. I think there are better ways to show this but this is the fastest with an reasonable result. Man.. have you really taken stats before? Does that data look normal to you? Look at the graph for fucks sake. You have a pretty severe right skew without a very defined bell! You can't use your deviation calculations if you don't establish normality. At least make note of the huge gaping inaccuracy here! You really need to leave detailed instructions of what each value means and how each one was calculated. Where exactly does each number come from? Where does 99% confidence come from? What do you mean by "-29"...? Lastly, if you don't now what tests you are running or if they are the correct, how the fuck can you make conclusions from them? I respect good statistical work, but the largest mistake in stats is to make claims that don't have their nuances clearly understood. | ||
|
ensign_lee
United States1178 Posts
 | ||
|
mechavoc
United States664 Posts
On July 11 2012 02:08 Stiluz wrote: It would be interesting if you could divide this by league too. Yes PLEASE would love to see the break out by league. I would bet that as skill increases so does balance | ||
|
lazyitachi
1043 Posts
On July 11 2012 23:31 lolcanoe wrote: <Rants> Welcome to TL standard of statistics. Don't hold your breath for correct application. Assumes shape is normal Player MMR can fluctuate min max by 1k MMR It still does not account for error on each individual player MMR | ||
|
skeldark
Germany2223 Posts
On July 11 2012 23:00 xian_ wrote: Best approach to meassure racial balance i came across. I'm reliefed to see that the differences seem to be relativly small. Hope you can gather enough data to calculate solid numbers for master+ soon. In the end that is the level most people here care about. I wonder wheter we will be able to see changes caused by future patches as well. Or even see which regions can adopt faster to changes. Keep it up. ps: If you expand your OP with information regarding the methods you might be able to avoid some discussions. Than the op would be 200 sites long. There will always missing information and there are always people who don't think and just assume what they want. I can not explain the hole mmr claculation and all basics about skillsystems in this post. If you know something about it you are able to ask reasonable questions. The rest.... Oh btw: @lolcanoe Based on my understanding of the MMR system, you would not expect normality... As a stats junky, I'm very suspicious of the statistics here. i hope you troll... | ||
|
Schafjaeger
Germany13 Posts
intelligent players like lasers -> players who like lasers pick protoss -> clever players are better than stupid players -> Protoss seems op User was warned for this post | ||
|
skeldark
Germany2223 Posts
Before you guys start the hole standard deviation question again, read throw this thread. Read the statistic discussion. Read it again. than think 15 min about everything. than read you own post again ... Back to the important points: -I dont think that only intelligent players like lasers! | ||
|
StarGalaxy
Germany744 Posts
Pls use red for Zerg and yellow for Protoss. I guess most of the ppl here are used to that bc of liquipedia. | ||
|
skeldark
Germany2223 Posts
On July 12 2012 01:10 OrbitalPlane wrote: On a side note to Not_That graphs: Pls use red for Zerg and yellow for Protoss. I guess most of the ppl here are used to that bc of liquipedia. already corrected. | ||
|
balosan
Poland232 Posts
On July 12 2012 01:10 OrbitalPlane wrote: On a side note to Not_That graphs: Pls use red for Zerg and yellow for Protoss. I guess most of the ppl here are used to that bc of liquipedia. Yea i agree they should change it, its big problem, how can i recognize races if COLORS are wrong. | ||
|
Not_That
287 Posts
On July 12 2012 01:10 OrbitalPlane wrote: On a side note to Not_That graphs: Pls use red for Zerg and yellow for Protoss. I guess most of the ppl here are used to that bc of liquipedia. Protoss icon is green on stream list, and that's the color I associate Protoss with based on my experience with TL. I imagine I'm not the only one. | ||
|
skeldark
Germany2223 Posts
On July 12 2012 01:19 Not_That wrote: Protoss icon is green on stream list, and that's the color I associate Protoss with based on my experience with TL. I imagine I'm not the only one. That is not correct. I use Teamliquid dark theme and for me its Teal! | ||
|
treekiller
United States236 Posts
I calculate the average skill of an race not the generell popularity! What is generell popularity? Is it like general popularity? | ||
|
skeldark
Germany2223 Posts
On July 12 2012 01:24 treekiller wrote: What is generell popularity? Is it like general popularity? No! The general of the popularity have nothing to do with this. | ||
|
Lysenko
Iceland2128 Posts
On July 11 2012 20:36 paralleluniverse wrote:] MMR should not be sampled. A sample of MMR isn't a set of independent observations like the height of a random group of people for which the usual techniques of statistical inference can be applied to.. Exactly. You can't apply the statistics of normal distributions to data sets whose points are not independent measurements. The problem isn't that it works in some cases and not others -- in fact the problem is that the more math you do with the wrong assumptions, the bigger your systematic error is in your result. I need to take a much closer look at how the OP is deriving his results. He asserts that correcting this didn't yield much change, but also it looks like he's not exactly doing a straight-up statistical analysis either, but instead comparing his results to some kind of monte carlo simulation... and of course those kinds of simulations require a whole different level of inspection to validate that I don't see having happened here. | ||
|
Lysenko
Iceland2128 Posts
On July 12 2012 00:08 lazyitachi wrote: Assumes shape is normal Player MMR can fluctuate min max by 1k MMR It still does not account for error on each individual player MMR Those last two are not necessarily problems with the analysis of a population, but the concerns of the guy you quoted are spot on. | ||
|
Lysenko
Iceland2128 Posts
No, he's right -- Elo, at least, usually follows a logistic distribution across a population. I pointed that out earlier, btw. http://en.wikipedia.org/wiki/Logistic_distribution | ||
|
1st_Panzer_Div.
United States621 Posts
On July 11 2012 23:31 lolcanoe wrote: Where the hell are you getting these numbers for? I'm assuming this 99.3 comes from some application of the normal distribution - have you been thorough established normality in MMR? Based on my understanding of the MMR system, you would not expect normality... As a stats junky, I'm very suspicious of the statistics here. Man.. have you really taken stats before? Does that data look normal to you? Look at the graph for fucks sake. You have a pretty severe right skew without a very defined bell! You can't use your deviation calculations if you don't establish normality. At least make note of the huge gaping inaccuracy here! You really need to leave detailed instructions of what each value means and how each one was calculated. Where exactly does each number come from? Where does 99% confidence come from? What do you mean by "-29"...? Lastly, if you don't now what tests you are running or if they are the correct, how the fuck can you make conclusions from them? I respect good statistical work, but the largest mistake in stats is to make claims that don't have their nuances clearly understood. Just give up, arguing stats on the TL is just worthless. Also there is definately a language barrier thing in this case. Overall a good experiment, I completely agree though that the execution and claims made here have no foundation. | ||
|
skeldark
Germany2223 Posts
On July 12 2012 01:53 Lysenko wrote: No, he's right -- Elo, at least, usually follows a logistic distribution across a population. I pointed that out earlier, btw. http://en.wikipedia.org/wiki/Logistic_distribution Beside all of this. People act like they have a problem with the way i analysed the data. But no one had one single point where i did something wrong. You can say: "i did not understand what you did at point X, can you please explain it and i allways did" You can say : i think you make an mistake at point Y because this will result in this error ... and i will check it. But what i see so far is this. -this is wrong because its wrong -Skill that dont take apm into account is no skill / you only look at win streaks you should check for the skill not the mmr - Lets talk about something total off topic about statistic... Often people start with so obvious wrong arguments that its hard for me to force myself to read the rest. I did not see an single post that point out an mistake. If you want to atack my analyse you are welcome, but you have to bring a point. And with you i dont mean you personal. Whatever mmr is, it have nothing to do with the analyse result! I think i show pretty simple that the "race-groups average" is to high to be a random mistake. And there is nothing in the data that is depending on the player race. So the race value must be biased. = not balanced with the whole data. If i just come up with a random number that i write behind each account and i accidental get such results on the first try every-time with different data i must be: A ) very very lucky or B) my random number is not random it shows an biased in the racedata. Yes im not not exactly doing a straight-up book statistical analysis. Thats what i said in the op from the beginning. But i do an statistical analyse. | ||
| ||
WardiTV Invitational
Playoffs
Creator vs CureLIVE!
Classic vs TBD
MaxPax vs TBD
[ Submit Event ] |
|
|
|
Replay Cast
RongYI Cup
herO vs Maru
uThermal 2v2 Circuit
Replay Cast
Wardi Open
Monday Night Weeklies
Sparkling Tuna Cup
The PondCast
|
|
|
EPT
