Deciding Map Popularity

[mibi]

I've been thinking about how one would measure the popularity of a map.

My first attempt just went by the number of pages it had in active games, but this was an inaccurate metric as world 2.1 games take longer to finish than doodle earth so there would be more active games. I also think the map movement stats thread is a bit in accurate as Doodle Earth games finish in 1/5 the time that World 2.1 games do leading to a higher total game count.

So I have devised this easy to use formula for measure a maps popularity.

Total number of active games
-------------------------------------------
Total number of territories

Thats it, just do the division and it does a good job or factoring game length into the equation.

Some sample stats.

Classic is 114.45
Doodle Earth is 18.83
Wordl 2.1 is 17.566
British Isles is 9.76
Feudal War 9.06

as you can see where as the map movement stats puts feudal war above world 2.1, this reveals a different story.

I think this is a more accurate metric as it involves game length. I don't really have time to do all the number crunching, but if someone wanted to give it a go, it would cool to compare data with the MMS.

[GrimReaper.]

or u could just use an anilitical program/alegrithim by using previous info as how many games were played when the map came out to the number of games played as of now on the maps(completed) also the number of terriorties/ rules/bounes/bombardment(ect.) can determine the pouplarity of the map before it even comes out to the beta mode.

this equation can also be set up to predice a players best map by factoring in his wins and loses turns not completed drops by using data from Random.org and even the analis of previous dice rolls can be factored in.


it seems like it would take forever but the dice rolls all must be on some kind of transcript (since lack pays them so he can use their engine.) and once everything is factored in a computer with greasemonkey or some kina of program(im not a computer savy{i like math}) it should not take long.


ut rember the more data the better.

[mibi]

or u could just use an anilitical program/alegrithim by using previous info as how many games were played when the map came out to the number of games played as of now on the maps(completed) also the number of terriorties/ rules/bounes/bombardment(ect.) can determine the pouplarity of the map before it even comes out to the beta mode.

this equation can also be set up to predice a players best map by factoring in his wins and loses turns not completed drops by using data from Random.org and even the analis of previous dice rolls can be factored in.


it seems like it would take forever but the dice rolls all must be on some kind of transcript (since lack pays them so he can use their engine.) and once everything is factored in a computer with greasemonkey or some kina of program(im not a computer savy{i like math}) it should not take long.


ut rember the more data the better.

uh...yeah. I prefer the k.i.s.s. principle.

[GrimReaper.]

yeah I'll explain it better tommorow it's 1 o'clock here in Chicago and I'm starting to get tired well good night all

[yeti_c]

Game length is a moot argument.

It only affects the stats for the first "average length of time of game on map" time period - then the stats will even back out.

C.

[MrBenn]

I'm not sure if it is a moot point... I guess most people carry a certain number of games; Freemiums only carry 4; the shorter the games they play, the more games they play.
I also expect that most Premiums try and carry what they deem to be an optimum number of games at any point (I struggle to keep up with games if I have too many active at once)...

[mibi]

Game length is a moot argument.

It only affects the stats for the first "average length of time of game on map" time period - then the stats will even back out.

C.

How is moot? When looking at active games, bigger maps will always fair better due to the amount of time it takes to complete a game.

[yeti_c]

I'm not sure if it is a moot point... I guess most people carry a certain number of games; Freemiums only carry 4; the shorter the games they play, the more games they play.
I also expect that most Premiums try and carry what they deem to be an optimum number of games at any point (I struggle to keep up with games if I have too many active at once)...

As Benn points out - people have a set number of games. - So just because a doodle earth game lasts half the length - than a World 2.1... - they'll (say) only have 1 slot at a time on either.

So once the initial rush of games is over - the games settle down into their average users for the map.

C.

[mibi]

I'm not sure if it is a moot point... I guess most people carry a certain number of games; Freemiums only carry 4; the shorter the games they play, the more games they play.
I also expect that most Premiums try and carry what they deem to be an optimum number of games at any point (I struggle to keep up with games if I have too many active at once)...

As Benn points out - people have a set number of games. - So just because a doodle earth game lasts half the length - than a World 2.1... - they'll (say) only have 1 slot at a time on either.

So once the initial rush of games is over - the games settle down into their average users for the map.

C.

Uh yeah, the shorter the games the more they play, which is why dividing by number of territories, which is a rough gauge of game length, will yield more accurate results if you are using active games as a metric. Whether or not a player is premium or not shouldn't make much of a difference.

[yeti_c]

I'm not sure if it is a moot point... I guess most people carry a certain number of games; Freemiums only carry 4; the shorter the games they play, the more games they play.
I also expect that most Premiums try and carry what they deem to be an optimum number of games at any point (I struggle to keep up with games if I have too many active at once)...

As Benn points out - people have a set number of games. - So just because a doodle earth game lasts half the length - than a World 2.1... - they'll (say) only have 1 slot at a time on either.

So once the initial rush of games is over - the games settle down into their average users for the map.

C.

Uh yeah, the shorter the games the more they play, which is why dividing by number of territories, which is a rough gauge of game length, will yield more accurate results if you are using active games as a metric. Whether or not a player is premium or not shouldn't make much of a difference.

That doesn't follow though... consider this...

One player has the following games. - They like to play 2 games at a time - when one finishes they start another.

Game A doodle Earth Started 1/1/08 Finished 1/2/08
Game B doodle Earth Started 2/2/08 Finished 1/3/08
Game C World 2.1 Started 1/1/08 Finished 1/3/08

So during that time - the active games for both maps is the same - completed games is higher for doodle earth
(Note how that can therefore skew the average towards doodle not the other way round).

C.

[mibi]

I'm not sure if it is a moot point... I guess most people carry a certain number of games; Freemiums only carry 4; the shorter the games they play, the more games they play.
I also expect that most Premiums try and carry what they deem to be an optimum number of games at any point (I struggle to keep up with games if I have too many active at once)...

As Benn points out - people have a set number of games. - So just because a doodle earth game lasts half the length - than a World 2.1... - they'll (say) only have 1 slot at a time on either.

So once the initial rush of games is over - the games settle down into their average users for the map.

C.

Uh yeah, the shorter the games the more they play, which is why dividing by number of territories, which is a rough gauge of game length, will yield more accurate results if you are using active games as a metric. Whether or not a player is premium or not shouldn't make much of a difference.

That doesn't follow though... consider this...

One player has the following games. - They like to play 2 games at a time - when one finishes they start another.

Game A doodle Earth Started 1/1/08 Finished 1/2/08
Game B doodle Earth Started 2/2/08 Finished 1/3/08
Game C World 2.1 Started 1/1/08 Finished 1/3/08

So during that time - the active games for both maps is the same - completed games is higher for doodle earth
(Note how that can therefore skew the average towards doodle not the other way round).

C.

Yes I get that, but you are not dividing the number of active games by the number of territories, which would then level it out.

[yeti_c]

Yes I get that, but you are not dividing the number of active games by the number of territories, which would then level it out.

Now - that would ruin Doodles stats completely despite the active games being the same.

If you were to divide "Completed" games by "number of territories" (or other metric to define length of game) then I could understand it...

Active games are self averaging after the inital burst of games...

C.

[yeti_c]

Yes I get that, but you are not dividing the number of active games by the number of territories, which would then level it out.

Now - that would ruin Doodles stats completely despite the active games being the same.

If you were to divide "Completed" games by "number of territories" (or other metric to define length of game) then I could understand it...

Active games are self averaging after the inital burst of games...

C.

Actually - thinking about that it would need to be a weighting for length of game - so a multiplication would be required - not a divide.

i.e. World 2.1 takes twice as long compared to Doodle - so :-
W2.1 Completed games * 2 = DE completed games

C.

[yeti_c]

Further suppositions.

Model 1.

Say
Map A has 5 games started every day. on average those games last 2 weeks.
Map B has 5 games started every day. on average those games last 1 week.

So day 1.
A = 5
B = 5

Day 2
A = 10
B = 10

Day 3
A = 15
B = 15

Day 4
A = 20
B = 20

Day 5
A = 25
B = 25

Day 6
A = 30
B = 30

Day 7
A = 35
B = 35

Day 8
A = 40
B = 35 (5 Completed)

Day 9
A = 45
B = 35 (10 Completed)

Day 10
A = 50
B = 35 (15 Completed)

Day 11
A = 55
B = 35 (20 Completed)

Day 12
A = 60
B = 35 (25 Completed)

Day 13
A = 65
B = 35 (30 Completed)

Day 14
A = 70
B = 35 (35 Completed)

Day 15
A = 70 (5 Completed)
B = 35 (40 Completed)

Day 16
A = 70 (10 Completed)
B = 35 (45 Completed)

Day 17
A = 70 (15 Completed)
B = 35 (50 Completed)
ETC...

The problem with this model - is it assumes that the "popularity" (games made per day) is constant. Once the first set of games finish - what do the players do - do they start another game? - If they like it they might? So if we say a 50% retention ratio... the model changes to this.

So day 1.
A = 5
B = 5

Day 2
A = 10
B = 10

Day 3
A = 15
B = 15

Day 4
A = 20
B = 20

Day 5
A = 25
B = 25

Day 6
A = 30
B = 30

Day 7
A = 35
B = 35

Day 8
A = 40
B = 38 (5 Completed) 3 restarted.

Day 9
A = 45
B = 40 (10 Completed) 5 restarted.

Day 10
A = 50
B = 43 (15 Completed) 8 restarted

Day 11
A = 55
B = 45 (20 Completed) 10 restarted

Day 12
A = 60
B = 48 (25 Completed) 13 restarted

Day 13
A = 65
B = 50 (30 Completed) 15 restarted

Day 14
A = 70
B = 53 (35 Completed) 18 restarted

Day 15
A = 73 (5 Completed) 3 restarted
B = 53 (40 Completed) 20 restarted. (3 previously restarted finished) 1 restarted again

Day 16
A = 75 (10 Completed) 5 restarted
B = 56 (45 Completed) 23 restarted (5 previously restarted finished) 3 restarted again

Day 17
A = 78 (15 Completed) 8 restarted
B = 57 (50 Completed) 25 restarted (8 previously restarted finished) 5 restarted again

Now the model degrades a bit slower again...

I've suddenly got bored and can't remember my point anymore either! Meh!!

C.

[DiM]

active games divided by number of terits is wrong.
why? simply because a map with a big number of terits doesn't imply a longer time to play. just look at conquest maps that have many terits but take few rounds to finish.

i'd say that there's no perfect formula for this but a pretty accurate one would be
completed games divided by availability time multiplied by average rounds per game.

for example.
map A has 1000 completed games and has been available for 100 days and the average round is 5. this means. (1000/100) * 5 = 50 points
map B has 500 completed games and has been available for 200 days and the average round is 30. this means. (500/200) * 30 = 70 points

thus making map B more popular (*)

more popular = people played it more.

then once this formula is applied stats need to be kept on a constant basis and fluctuations in the score have to be recorded. the map with the highest momentum is the most popular at that moment. the map with the highest overall score is the most popular of all time.



PS: or you could simply say classic is the most popular and end the debate

[GrimReaper.]

or u could just use an anilitical program/alegrithim by using previous info as how many games were played when the map came out to the number of games played as of now on the maps(completed) also the number of terriorties/ rules/bounes/bombardment(ect.) can determine the pouplarity of the map before it even comes out to the beta mode.

this equation can also be set up to predice a players best map by factoring in his wins and loses turns not completed drops by using data from Random.org and even the analis of previous dice rolls can be factored in.


it seems like it would take forever but the dice rolls all must be on some kind of transcript (since lack pays them so he can use their engine.) and once everything is factored in a computer with greasemonkey or some kina of program(im not a computer savy{i like math}) it should not take long.


ut rember the more data the better.

uh...yeah. I prefer the k.i.s.s. principle.

the analitical program can be modified were it can tell a players most efficient maps/ settings

[barterer2002]

Is there a way to measure unique players to a map?

Its seems to me that a fundamental part of the original idea is that you've got two players. One prefers World 2.1 and the other prefers Doodle. The Doodle player can hit 10 games in the time it takes a World 2.1 player to do 1 (numbers are not confined to facts in this example). Thus, if we're looking at games by the number of games played then Doodle appears to be more popular the question is whether this is so.

If there were a method to create a formula that would count the number of unique players on a map and the number of games on a map it might show popularity better.

[GrimReaper.]

ill give it a try

[hulmey]

good try mibi, but im afraid AoR is a prime example why its incorrect. Even 4 player games finish in record time on this map and it has a huge number of terits

[mibi]

good try mibi, but im afraid AoR is a prime example why its incorrect. Even 4 player games finish in record time on this map and it has a huge number of terits

I know, it doesn't work for conquest maps.

[hulmey]

good try mibi, but im afraid AoR is a prime example why its incorrect. Even 4 player games finish in record time on this map and it has a huge number of terits

I know, it doesn't work for conquest maps.

not for all conquest maps. I'm running a New Worlds Tournament and i think they are in Round 80...Its just a few like AoR and Feudel