Forged Alliance Forever

Re: Training a neural network to predict results of setons g

2016-03-21T23:23:01+02:00

sorry for derailing but

ODIN WHERE HAVE YOU BEEN

Statistics: Posted by TheKoopa — 21 Mar 2016, 23:23

Re: Training a neural network to predict results of setons g

2016-03-21T17:02:32+02:00

Omfg the link you showed is really amazing, I didn't know this before and it's amazing, I really love it, even if it's not really easy to make it (but not too hard), thanks you very much for this discovery !

Statistics: Posted by odin002 — 21 Mar 2016, 17:02

Re: Training a neural network to predict results of setons g

2016-01-22T00:58:36+02:00

Hey ckitching that dropbox link doesn't work anymore. Any chance of getting another database dump?

Statistics: Posted by Axle — 22 Jan 2016, 00:58

Re: Training a neural network to predict results of setons g

2015-08-09T05:53:44+02:00

lie < huge lie < stats.

Statistics: Posted by justmakenewgame — 09 Aug 2015, 05:53

Re: Training a neural network to predict results of setons g

2015-08-11T19:11:00+02:00

I have some very preliminary results that I'd like to compare to what TrueSkill gives us, so that we know how much accuracy can be expected. Any ideas on how to do that? I would not be too thrilled though, atm the best architecture is correct only about 61% of the time.

Statistics: Posted by Memto — 09 Aug 2015, 01:25

Re: Training a neural network to predict results of setons g

2015-06-16T13:19:07+02:00

I'm also looking at the data, but I'm more interested in which maps are played how often, the most popular size of maps, and if certain people play better on certain sized maps.

Unfortunately, that data dump seems to be missing the gameID for it's entries, which makes handling it a little difficult.

Statistics: Posted by ZenTractor — 16 Jun 2015, 13:19

Re: Training a neural network to predict results of setons g

2015-06-16T09:31:11+02:00

Nombringer wrote:
Any update on this? I'm pretty interested

I've done some reading and have decided on some architectures that I think I'm going to try. I've also done some preprocessing of the data, to keep the training algorithm separate from the data format. I have a job taking my time, so getting this project done might take a while. At the moment I hope to have this finished by fall.

da_monstr wrote:
Out of curiosity : are you using a custom application (if so, what language are you using) or are you using Matlab?

I'll do the input processing in C# and use a Matlab macro for training. I'm not sure what the rules are about using Matlab's neural network toolbox, so I may have to write my own implementation.

Statistics: Posted by Memto — 16 Jun 2015, 09:31

Re: Training a neural network to predict results of setons g

2015-05-28T11:50:00+02:00

Any update on this? I'm pretty interested

Statistics: Posted by Nombringer — 28 May 2015, 11:50

Re: Training a neural network to predict results of setons g

2015-05-21T02:55:07+02:00

[youtube]https://www.youtube.com/watch?v=bxe2T-V8XRs[/youtube]

This is a pretty good into to NN for those who are unfamiliar.

Statistics: Posted by Nombringer — 21 May 2015, 02:55

Re: Training a neural network to predict results of setons g

2015-05-18T20:40:12+02:00

Here is the complete dump (minus the most recent couple of months):
https://www.dropbox.com/s/mukkqlx15s8ll ... sv.gz?dl=0

Enjoy.

Statistics: Posted by ckitching — 18 May 2015, 20:40

Re: Training a neural network to predict results of setons g

2015-05-18T20:12:08+02:00

Ah.

Sheeo just informed me I missed a spot.

And by "a spot" I mean "Everything before mid-2014".

There will be a brief interlude while the several GB I missed are extracted

Statistics: Posted by ckitching — 18 May 2015, 20:12

Re: Training a neural network to predict results of setons g

2015-05-18T19:29:19+02:00

Oh, I do beg your pardon, I forgot to post the explanatory accompaniment:

http://sprunge.us/LPcc

The datafile dumped is the output of:

Code:: SELECT login.login as player, b.game_type as game_type, game_player_stats.faction, game_player_stats.color, game_player_stats.team, game_player_stats.place, game_player_stats.mean, game_player_stats.deviation, game_player_stats.after_mean, game_player_stats.after_deviation, game_player_stats.score, game_player_stats.scoreTime, b.start_time start_time, b.end_time as end_time, b.map_name as map_name, b.map_file as map_file FROM game_player_stats INNER JOIN login ON game_player_stats.playerId = login.id INNER JOIN ( SELECT game_stats.id, game_stats.mapId, game_stats.startTime start_time, game_stats.EndTime as end_time, game_stats.gameType as game_type, table_map.name as map_name, table_map.filename as map_file FROM game_stats INNER JOIN table_map ON table_map.id = game_stats.mapId ) as b ON game_player_stats.gameId = b.id

The fields in the output are separated by tab characters. The records are separated by newlines. The first record is the column names.

This format would be suitable for import into prettymuch any database system you might have, and spreadsheet software will probably happily take it as well (if it weren't so enormous, anyway). If you only want a narrow portion of it I suggest you use `grep` or some similar tool to filter it down to a more manageable size before having your way with it.

If you care about game count, notice that that field is implicit in the data set and must be determined by counting records for each player in chronological order. Also note that datestamps are sometimes missing, because ZeP, so your should probably use ID numbers as Lamport clocks if you're going to try and reconstruct that.

Statistics: Posted by ckitching — 18 May 2015, 19:29

Re: Training a neural network to predict results of setons g

2015-05-18T18:50:30+02:00

What is the best way to read and sort this data? Where do you see all the players and their spots and where does it list who won?

Statistics: Posted by Ionic — 18 May 2015, 18:50

Re: Training a neural network to predict results of setons g

2015-05-18T16:16:44+02:00

In case it's of interest to anyone else, here is the database dump:

https://www.dropbox.com/s/7wtqe7zgkui99 ... sv.gz?dl=0

Go forth and built neural networks and suchlike. Or something else. This represents a few years worth of game history.

Statistics: Posted by ckitching — 18 May 2015, 16:16

Re: Training a neural network to predict results of setons g

2015-05-15T17:05:33+02:00

ZLO_RD wrote:
That is pretty fun idea... But if you just make 4 separated ratings for every player, back front beach and rock and add the importance of spot ...

Front=1
Beach/Rock=5
Back=9.99E99

I suppose this is correct

Statistics: Posted by zeroAPM — 15 May 2015, 17:05