Administrative Efficiency 61.94 %

If it is an average, it should not ignore the top or bottom halves. It should include all those who are actively playing.

That is called a bias statistic.

melonhead wrote:If it is an average, it should not ignore the top or bottom halves. It should include all those who are actively playing.

That is called a bias statistic.

Esactly but hey its nothing new Admin is BIASED and he openly admins it so either accept you play a game where Admin dictates how you play and has a tendensy to help those he likes bette than others or don't.

It ignores the outliers. Makes it less influenced by extremes. Like 1 person with a 60M army size whereas number 2 has a 10M army size and a gausse-curve distribution from there. Idea is chop off the extremes (both extremely large and extremely small) and you get a representative picture of the account size of normal players.
There are other ways of doing this, but I think there is not much wrong with this approach.
@Nimras: There is no need for this anger. He already changed it to something more reasonable, because in a way he agreed with you, but for some reason or another some people are still experiencing problems. I am sure it will work out eventually.

Jiro wrote:It ignores the outliers. Makes it less influenced by extremes. Like 1 person with a 60M army size whereas number 2 has a 10M army size and a gausse-curve distribution from there. Idea is chop off the extremes (both extremely large and extremely small) and you get a representative picture of the account size of normal players.
There are other ways of doing this, but I think there is not much wrong with this approach.
@Nimras: There is no need for this anger. He already changed it to something more reasonable, because in a way he agreed with you, but for some reason or another some people are still experiencing problems. I am sure it will work out eventually.

Ohh me angry no i just poste what everyone who has followed his behavour lately knows.

He has proven it a few times. But yes some of us can't get AE to update it self.

Nimras wrote:

melonhead wrote:If it is an average, it should not ignore the top or bottom halves. It should include all those who are actively playing.

That is called a bias statistic.

Esactly but hey its nothing new Admin is BIASED and he openly admins it so either accept you play a game where Admin dictates how you play and has a tendensy to help those he likes bette than others or don't.

if you have little or no clue about statistics then dont talk crap but instead ask why things are done the way they are done.

there is no such thing as a bias statistic.
However there is such a thing as a robust and non robust statistic

http://en.wikipedia.org/wiki/Interquartile_range
I am using a robust statistic so that it ignores extreme values which have no significant representation in the sample as jiro already has explained.

Admin wrote:

Nimras wrote:

melonhead wrote:If it is an average, it should not ignore the top or bottom halves. It should include all those who are actively playing.

That is called a bias statistic.

Esactly but hey its nothing new Admin is BIASED and he openly admins it so either accept you play a game where Admin dictates how you play and has a tendensy to help those he likes bette than others or don't.

if you have little or no clue about statistics then dont talk crap but instead ask why things are done the way they are done.

there is no such thing as a bias statistic.
However there is such a thing as a robust and non robust statistic

http://en.wikipedia.org/wiki/Interquartile_range
I am using a robust statistic so that it ignores extreme values which have no significant representation in the sample as jiro already has explained.

Lol i weren't talking about statistic actually admin and again you move to petty insults as a little child your so full of your self aren't you. You owe me for 24 hours BAN because of your childish behavour and petty insults i kinda think i deserve that .

You just proved you don't read my post proberly lol I don't care about his statistic i care about the fact he said biased .

hence proving my point

Admin wrote:hence proving my point

Lol admin tsk tsk has to sink this love ohh well.

Niamras cool it, I have only seen undue crap. I know Admin has issues with the way some play and get ahead of others. Which the main reasons behind it I can agree with. however it changes the status quo, which is what everyone has issues with.

seaborgium wrote:Niamras cool it, I have only seen undue crap. I know Admin has issues with the way some play and get ahead of others. Which the main reasons behind it I can agree with. however it changes the status quo, which is what everyone has issues with.

Lol he is Biased and he is double standards and heck i even say it worse than Jason who at least admits it and don't care lol.

Lord Ishurue wrote:

im back to 25% again .

guys of 6 mil army with 70% military are making more income .

Lord_Ishurue Kyora 26,815,600 158,809,251 25.13%

Lord Ishurue wrote:
???? Kyora 5,796,610 38.35%

????????? 10,056,000 61.97%

how is this correct ?

i have 25% AE because i am the 3rd largest army ingame ? thats BS .

Lors Ishuru may i surgest looking in the bug sections it took mine hours to update after admin fixed it and my officer who is 2,7mill haven't even updated she tried as well.

Admin either can't find the reason for this bug or just like to show biase ness ohh and yes Admin for your ban on me and insults i will have fun with giving back :p.

Ohh and don't think i am angry or have a thing for you i just injoy this to muuch to stop

Lord Ishurue wrote:
???? Kyora 5,796,610 38.35%

????????? 10,056,000 61.97%

how is this correct ?

i have 25% AE because i am the 3rd largest army ingame ? thats BS .

found the issue now

to think my question on why it happened started this. admin if you are basing on 95% band or some other statistical method, its not an average of army size you are getting rather a mean of weighted statistics. I really didn't care that it dropped, just the why lol

Nimras wrote:

Kenzu wrote:Administrative efficiency dropped so much because 170 accounts registered today. That's all.

AE is based on active players, yes. Since only a minority of users who register keep playing on the 2nd day the AE should fall back.

btw, do you have any idea why people register but most of them don't bother logging in a second time?
This is something which exists in each single game, you can always see that a game has tousands of registered players but probably only 10% active players or less.
If we ignore those people who played a long time and focus only on the people who quit playing the day they registered. Why do you think it is this way?

Kenzu no offense i give a crap ass about it what i want is for ADMIN to make sure when so many accounts are made it doese not affect AE this much i already feel screwed by the dear admin once and he still owe me the losses for that, And i want compensation for the losses now because of a system thats to easy to affect.

I don't like your attitude.

Besides, AE has been made for a reason and it does it job well.

stars wrote:Do you know that if the entire server made 1 extra account we would all drop the average to so low AE would be 5%? I feel that it may be flawed. I believe you should change the average to the sum of maybe the top 100 army sizes to make it fair to your players who STARTED FROM THE BEGINNING. It seems the only thing that is being done is that everyone who grows larger always gets punished for those new recruits they bring in. How likely would your old players want to bring in new players if all we get is punishment to our income and ETC.

There is a restriction how low AE can be. There can be even 1.000.000 registrations on one day and AE will still not reach 5%.

Jiro wrote:Admin, do you think there is a way you could make global average army size fluctuate less?
Or is the current volatility of average armysize intentional?
If it is unintentional, I can imagine something like making the current effective global average armysize the average of the GAA of the past 7 days. That would have made the dip from today much less. There would be a steady decline if lots of new players joined and stayed, but only a small dip if they left again.
And yes, my AE went from 60% to 33%, so it is inconvenient. Then again, the game is not meant to be convenient.
As for an AE of 5%, I thought it approached 20% as your armysize approaches infinity and there is no way you'd get to 5%.

The idea of looking at average armysize over the last 7 days is definitely a good idea. I can imagine it be put in action. It will decrease AE fluctuations.

Kingofshinobis1 wrote:

Admin wrote:It looks at who's been active in the last day, ignores a quarter of the smallest and biggest players and then finds out the average of the rest.

Well obviously the system is a little flawed in judging the active accs.

I am not sure why it "ignores" some players. Shouldn't it take ALL active players into account? Doesn't make much sense but that's just my opinion. (I guess its to regulate AE but idk for sure).

I don't really have any suggestions per say for the AE but I will say this is a little bit ridiculous.

I also think it should not ignore any players. Afterall if you consider all players there will be less fluctuations.

stars wrote:Do you know that if the entire server made 1 extra account we would all drop the average to so low AE would be 5%? I feel that it may be flawed.

Not only would such system be totally flawed...
What would interest me more is that how you came up with that ridiculous idea that the system works in such a crappy way.

Though I am sure you have your reasons for doing things this way, and I have no issue as such.

There is most definitely such a thing as a bias statistic. It's a fundamental for surveyors and accountants. Being the latter I know this... but if you wish to quote the all mighty wiki


http://en.wikipedia.org/wiki/Bias_%28statistics%29

In statistics, the term bias is used for describing several different concepts:
A biased sample, sometimes classified as a result of selection bias, is one in which some members of the population are more likely to be included than others.

The bias of an estimator is the difference between an estimator's expectation and the true value of the parameter being estimated.

* Omitted-variable bias is the bias that appears in estimates of parameters in a regression analysis when the assumed specification is incorrect, in that it omits an independent variable that should be in the model.


just saying...

Kenzu wrote:

Nimras wrote:

Kenzu wrote:Administrative efficiency dropped so much because 170 accounts registered today. That's all.

AE is based on active players, yes. Since only a minority of users who register keep playing on the 2nd day the AE should fall back.

btw, do you have any idea why people register but most of them don't bother logging in a second time?
This is something which exists in each single game, you can always see that a game has tousands of registered players but probably only 10% active players or less.
If we ignore those people who played a long time and focus only on the people who quit playing the day they registered. Why do you think it is this way?

Kenzu no offense i give a crap ass about it what i want is for ADMIN to make sure when so many accounts are made it doese not affect AE this much i already feel screwed by the dear admin once and he still owe me the losses for that, And i want compensation for the losses now because of a system thats to easy to affect.

I don't like your attitude.

Besides, AE has been made for a reason and it does it job well.

Kenzu.

1. My attitude is well deserved both for admin and you so stop whining as you seem to do that.

2. AE was made for a reason yes i know the same reason Plauge was made in GW when someone got to high or same way as growth in Ascended slow down when you reach a size so please.

3. The way AE worked before was flawed and it still is. the fact that Admin do not take in count every single ACTIVE account makes it not really giving what it should. Plus the fact IT took every new account made as well into the option means someone who wanted to have fun could put so many accounts in that every single user almost got to 20% AE in no time.

Another flaw tho this is fixed almost it can still be misused. In this case i will say GW wins the battle as the way that works do have a advantage if we forget how many plays and how big it is would what happen this time and how this is calculated not be able to happen at any point.

Anyway i am looking forward to see how the new changes in the AE will work you for the game.

At least AE didn't screw everyone over so much as the TRADE RATIO bullshit did lol worst update EVER .

melonhead wrote:Though I am sure you have your reasons for doing things this way, and I have no issue as such.

There is most definitely such a thing as a bias statistic. It's a fundamental for surveyors and accountants. Being the latter I know this... but if you wish to quote the all mighty wiki

http://en.wikipedia.org/wiki/Bias_%28statistics%29

There is such a thing as bias IN statistics, there is no such thing as a bias statistic.
The results of a statistic can be biased, because of the conditions it has been done in.
There is sample bias...
plenty other examples

The interquartile range calculation is NOT a "bias statistic" whatever that is.

The interquartile range has no bias in that it avoids extreme values by only considering the 50% of all values that are closest to the median
That is called a robust test

compared to the non-robust test nimras is suggesting to use which is prone to errors and misrepresentation

OK, sorry, I can't make head nor tails from how AE works currently. I thought it updated once a day, at midnight gametime + when you clicked on the income link.
I had a 177M per turn income earlier today and could fully afford to pay al my officers and now I have a 146M income and I can only afford a severely reduced rate. What is going on? I am lost.
On a side note, I gather from melonhead's post that the interquartile average can be biased if the implicit assumption (that the middle range is a representative group) is false. That would be the case if the distribution is severely skewed, like when there are many more new accounts compared to active old accounts. If the new accounts outnumber the old ones by 4:1, they'd get to determine the complete picture. I think the update to count only accounts that are played for over a day helps eliminate this problem, but every statistic has its own inherent weakness.