Massive Assault
http://www.massiveassaultnetwork.com/forum/

Data used in Battle and creating a database
http://www.massiveassaultnetwork.com/forum/viewtopic.php?f=30&t=1776
Page 1 of 1

Author:  warchrisMA [ Tue Oct 19, 2004 9:58 am ]
Post subject:  Data used in Battle and creating a database

hello,


i have a look on differents datas (.dat) and way they are used (for instance in glyco forms.)

i do think it may be very usefull to use a Mysql database to store datas.
it would be easier to manipulate.

would it be possible to make a small database (you got one because you put phpbb on one) dedicated to clan war in a first step but with whole data.

i've thought of a MCD (if someone want to tell me his opinion on it, i can give in a model), i'm using DataBase Designor 4.

if your are not sure this is a good idea i will make one on rotm website and show data in order for you to have an idea.


i got some good idea (ex : different way to show battle results with graphics : gd library or flash/php grahics)

the actual glyco doesn't show many stats about number of battle win/loose with PL/FNU witch is an important indicator.

and it's acurate only with many battles.

Author:  Maelstrom [ Tue Oct 19, 2004 10:17 am ]
Post subject:  Re: Data used in Battle and creating a database

warchrisMA wrote:
the actual glyco doesn't show many stats about number of battle win/loose with PL/FNU witch is an important indicator.

and it's acurate only with many battles.


Actually, we are using a modified version of Glicko that does keep track of PL vs FNU victories on a per map basis. These values are then averaged together to form our modified values.

Here are some excerpts from the Clan War rules:

"
i) The Glicko rating will use the Averaged Glicko approach as shown at
http://www.massiveassault.com/clans/nwo/allratings2.php .

ii) This rating will be modified only to include the following maps in the average: Emerald, Antarcticus, Craetus, Noble Rust, Anubis, and New Paradise.
"

The averaged Glicko does a better job of dividing games into comparable sets, but due to its averaging, it requires even more complete games to be accurate.

Author:  warchrisMA [ Tue Oct 19, 2004 11:16 am ]
Post subject: 

i didn't mean, glyco wasn't good ('cause i was talking of glyco2...)

i meant that a player can't easily answer to those simple question :

how many battle i've player with FNU or PL ?
how many battle i've played with player x ?

how many battle i've won with players with glyco rating higher than 1800 on crateus.

the only way is to read 100 lines and counting....

Author:  Silicor [ Tue Oct 19, 2004 11:38 am ]
Post subject: 

Speaking of the glicko, I think we should require a minimum number of games played before entering clan wars. There is a dramatic difference in the number of units you start with compairing the best and worst player. A player who has not established their glicko score will unbalance the wars by making their clan either very strong or very weak.

This does not stop anyone from entering, just makes them play some games before. Their clan mates could challenge them and establish a statistically sound glicko pretty fast.

Author:  warchrisMA [ Tue Oct 19, 2004 11:44 am ]
Post subject: 

i agree with you but it is very difficult to say how many battle to have done to be acurate.

if you consider 100 battles for each map...
you got many battles and may be only Marshall...

if you say 10 battles each MAP (near 100) : it's not realy acurate...

Author:  Silicor [ Tue Oct 19, 2004 11:52 am ]
Post subject: 

I don't think we need to be crazy about this. Lets say 30 total games on the maps(not on each map) used to put together the average would be enough. The big fear is a new player lik Madjazz comes around who had few games before the world champioship tourney. His glicko rating put him in the 300s, where after all the tourney games he is 33 and from what I saw will continue to improve. Him playing at the beginning of his career would have gave his team a large advantage in units. He was in the 30s after around 30 games. For that matter, I did the same thing. I started ranked very low, but advanced quickly based on my past wargame experience.

We don't need to be perfect here since prices are adjusted over time, but we should have folks starting the wars with glickos in the right order of magnitude.

Author:  warchrisMA [ Thu Oct 21, 2004 10:38 am ]
Post subject:  data's have been imported

well, i got the feeling that data aren't complete.

i got coutries data but not all gif linked to those countries (the glyco takes gif on a part of the MAN server i can't see, i could do the same but it would no be efficient... so i try to get gif i need with glyco table...

see pb on rotm menu player list.
i got too many player (9600..) it's because many people (like me) has got 2 or 3 nickname.

that not the main pb.
for battles, i got near 30 000 , and battles should be near 44 000...
so :
1) programmer of Man has made counter begin near 10 000 to make people think many battles already played..
2) 10 000 battled of test that has been erased.
3) my data import isn't good (might be , i don't think so)
4) other reason

it can be several reason of those.


for 1st test it's gonna be ok.

Author:  Artanis [ Thu Oct 21, 2004 10:44 am ]
Post subject: 

Well, #1 is not a possibility, because people were around and playing long before I was, and my first game was #4500

On #2, it might have erased broken games as well, though I don't like the thought of there being 10,000 of them.....

Author:  MortonHQ [ Thu Oct 21, 2004 2:23 pm ]
Post subject: 

Is the rating for the war the total rating or just the rating on games played against the other players in the war.

It might be more useful if it was just the relevant games, but I guess that might not give enough games to rate each player on.

Author:  Artanis [ Thu Oct 21, 2004 2:59 pm ]
Post subject: 

Total rating.

Author:  Enforcer [ Thu Oct 21, 2004 3:36 pm ]
Post subject: 

we might consider just using clan war ratings once we have some data to use :) (so not for a few wars)

Author:  Maelstrom [ Thu Oct 21, 2004 8:12 pm ]
Post subject: 

The problem with any kind of rating system is the amount of data we need to make an accurate rating. Its a particular problem with Massive Assault, as each game takes weeks to complete on average, and so we have so little data to work with from the beginning. Especially considering that battles on different planets have different levels of tactics and strategy, and are hard to group into one rating accurately.

MWigor uses a special rating system for Tournament seeding, which looks at glicko ratings, MA ratings, and performance in previous tournaments. With this amount of data to draw from it could possibly be more accurate than Glicko alone (especially since it looks at tournament results), but it will be difficult to make this rating easy to calculate for large numbers of people. Once the method is given a more digestible form, it may be a better solution.

Author:  Norat [ Sun Oct 24, 2004 10:29 am ]
Post subject:  Re: data's have been imported

warchrisMA wrote:
i got coutries data but not all gif linked to those countries (the glyco takes gif on a part of the MAN server i can't see, i could do the same but it would no be efficient... so i try to get gif i need with glyco table...

Yes, not all countries have their gifs, even on MAN web server. Maybe I can help you on this matter. Contact me by email (vzverev@nm.ru) if you interested.

Page 1 of 1 All times are UTC - 5 hours
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/