Statistics

jiriki
176
old people
Admins
Provider & Webmaster
Posts
490
Location
Oulu, Finland
Joined
1 May 2009
28 September 2009 - 13:16 CEST
#1

There're about 5000 rounds in the database now. These can used to generate lots of interesting statistics, but it takes quite a bit of time (implementing is rather trivial though), so I'd like to know what kind of stats you would like to see. All kinds of ideas are worth mentioning, but you have to remember that log files don't have full information, they don't show if a research completed or welding requests etc.

Get to the spaceship!
Elvisq
674
Team Four
Posts
57
Location
Bielsko-Biala, Poland
Joined
9 October 2005
29 September 2009 - 14:37 CEST
#2

What would i like to find out (from personal statistic ) might be the avarage lifetime of liveforms, maybe amout of rts bitten/round,

from team one, time when reserches starts (avarage etc), time of hivedrop etc.

Tane
91
HasBeens & NeverWas
Posts
57
Location
Jyväskylä, Finland
Joined
7 May 2005
4 October 2009 - 14:50 CEST
#3

Kills and Deaths

KPD and KPM

Stats from structure destroying and building

How many ammoback, medpack, weapons and structure dropped by commander

dugi
203
Donors
Server Donator
Posts
222
Location
Madrid, Kazakhstan
Joined
9 May 2005
4 October 2009 - 17:00 CEST
#4

What weapons did you use to kill someone, time spent playing (after readying up), number of consecutive frags without dying and number of deaths without killing too ;D.

 

;D
GibbZ
Noavatar
9L
Gather Moderators
NS1 Gather Mod
Posts
154
Location
Sierra Leone
Joined
7 May 2005
4 October 2009 - 18:04 CEST
#5

I'm more interested in player roles. i.e. time spent fading, time spent lerking and also buildings destroyed/created stats aswell, so you can tell the good teamplayers from the frag whores.

Fana
Noavatar
Archaea
Donors
Movie Donator
Posts
291
Location
Oslo, Norway
Joined
6 May 2005
4 October 2009 - 19:12 CEST
#6

I'm not a big fan of player KPD statistics, it's just going to promote frag whoring.

#archaea @ irc.quakenet.org

Tane
91
HasBeens & NeverWas
Posts
57
Location
Jyväskylä, Finland
Joined
7 May 2005
4 October 2009 - 20:00 CEST
#7

In the end KPM is the best way to see how good player is. I haven't ever played with person who would play for frags so I don't see how it would encourage anyone to be fragwhore. I would also see it especially practical when recruiting new players.

You can't really count advanced teamplay(e.g. blocking fades, covering, positioning) so don't know how could you see from stats who is teamplayer and who's not.

 

Fana
Noavatar
Archaea
Donors
Movie Donator
Posts
291
Location
Oslo, Norway
Joined
6 May 2005
4 October 2009 - 21:33 CEST
#8

"In the end KPM is the best way to see how good player is."

No, it isn't, there are far too many variables. A player who plays first Fade is going to rack up a ridiculous KPM and KDM compared to the Skulk specialists, but that doesn't make him a better player. A player on a good team is going to get a lot better KPM and KDM than a player on a poor team, even if the player on the poor team is individually a better player than the player on the good team. The list goes on.

 

It won't necessarily make people start hunting for frags (at least not any more than they do right now), but it will consciously or sub-consciously influence people, making them think frags are more important than they should be. We don't need more focus on frags.

#archaea @ irc.quakenet.org

sublime
5
Posts
488
Location
United Kingdom of Great Britain and Northern Ireland
Joined
6 May 2005
4 October 2009 - 22:01 CEST
#9

Time spent on the server NOT playing. Nice tie-in with your "what's wrong with ns1" moanpost. Average it over clans perhaps and see which are the greatest time wasters. Won't be perfect because of the fact you get clans waiting on other clans' players, but could be interesting.

Tane
91
HasBeens & NeverWas
Posts
57
Location
Jyväskylä, Finland
Joined
7 May 2005
4 October 2009 - 22:35 CEST
#10

Well make different KPD and KPM for different roles. I know this won't work best for aliens but for marines I don't really see reason why wouldn't it tell something about the player. One could argue that pressure marines get better KPM but on other hand they normally get worser KPD. Also stats would show in which division player has played so you can easily see how good player really is. Still I'm not arguing that you would know everyting about player from stats but it would help to track some potential players.

GibbZ
Noavatar
9L
Gather Moderators
NS1 Gather Mod
Posts
154
Location
Sierra Leone
Joined
7 May 2005
4 October 2009 - 22:53 CEST
#11

Marines: Early shotgun?

I mean how can you rely on a KPM result. If you have a rather quick game a good shotgunner (not automatically a great marine) can outscore his team even if other players are just as good. I mean really, NS is too complicated to rely on KPM.

You have cappers, pressure teams, base guards, commanders... ect...

The main shotgun in a team should outscore the cappers heavily really.

Voodo
604
Two Team Squad
Posts
55
Location
Budapest, Hungary
Joined
29 September 2005
4 October 2009 - 23:07 CEST
#12
make a formula, that includes win/lost rounds, k/d ratio, matches played, and how old is the player, etc
Bombing for peace is like fucking for virginity.
Voodo was here.
Tane
91
HasBeens & NeverWas
Posts
57
Location
Jyväskylä, Finland
Joined
7 May 2005
5 October 2009 - 11:01 CEST
#13

Early shotgun doesn't really make KPM invalid. People play in different roles and middle-late game evens the starting roles. People don't play in same roles as marine every time which also evens KPM. I like to remind again that one shouldn't just look someone's stats without looking the big picture. For example if watch Green's stats I know that he is often early sg and flash fade. This information should be also to be seen in stats.

I'm probably thinking this case from different angle than you. I don't support that you make list who has best KPM and KPD if that is what you fear. I'm thinking situation were high-level clan is recruiting players from lower-level clans. You get 50 players and you can't just put them all to trial. I just used to pick those "lucky" ones by their reputation. For example I didn't even give shot to enigmatic back in 3.0. By looking stats and watching few demos you could get enough information to decide who get chance to trial.

sublime
5
Posts
488
Location
United Kingdom of Great Britain and Northern Ireland
Joined
6 May 2005
5 October 2009 - 19:26 CEST
#14

That would be largely dependent on the quality of the players team. You would need a very large community for there to be enough finely balanced games for that analysis to work properly.

lump
81
Posts
499
Location
United Kingdom of Great Britain and Northern Ireland
Joined
7 May 2005
6 October 2009 - 14:21 CEST
#15
Stats in ns are just a complete load of bullshit, I quite happily die to hit a hive so that my team can come through and clear the pg builders..  This makes me a worse player despite it having a massive and direct impact on the game result?  Pssh..  nevermind all the other shit.   I don't even want to count how many people will take an easy frag dancing around with their teammate instead of go for the hard frag coming at their teammate from behind.  AKA.. cover... lol..
Voodo
604
Two Team Squad
Posts
55
Location
Budapest, Hungary
Joined
29 September 2005
6 October 2009 - 14:23 CEST
#16
just make it won rounds/all rounds
Bombing for peace is like fucking for virginity.
Voodo was here.
Tweadle
Noavatar
Gather Moderators
NS1 Gather Mod
Posts
159
Location
South Africa
Joined
10 May 2005
6 October 2009 - 14:36 CEST
#17

Time spent for teams to ready up please. Then I can use the statistics constructively and never play slow teams.

sublime
5
Posts
488
Location
United Kingdom of Great Britain and Northern Ireland
Joined
6 May 2005
6 October 2009 - 15:24 CEST
#18

Like I said you can't always blame the team who spent the longest on the server...

Tweadle
Noavatar
Gather Moderators
NS1 Gather Mod
Posts
159
Location
South Africa
Joined
10 May 2005
7 October 2009 - 12:56 CEST
#19

Time spent for teams when there's six on each team to ready up please. Then I can use the statistics constructively and never play slow teams.

lump
81
Posts
499
Location
United Kingdom of Great Britain and Northern Ireland
Joined
7 May 2005
7 October 2009 - 14:10 CEST
#20
lol you can just play the teams that take 30 mins to get their sixth... or not play, as the case may be.
Tweadle
Noavatar
Gather Moderators
NS1 Gather Mod
Posts
159
Location
South Africa
Joined
10 May 2005
7 October 2009 - 14:30 CEST
#21
...or do the complete opposite.
lump
81
Posts
499
Location
United Kingdom of Great Britain and Northern Ireland
Joined
7 May 2005
8 October 2009 - 06:07 CEST
#22
well you'll need the time spent for the team when there's 5 then wont you! lol
Miglecz++
1643
Posts
42
Location
Debrecen, Hungary
Joined
7 February 2008
8 October 2009 - 09:10 CEST
#23

I am interested in personal statistics so everybody can improve playing style/skill.

ex.: Resources spent/gathered by players.

Marines:

  • medpack/ammo/catalyst/weapons/jp/heavy/welder etc. items drop on death counts as minus res.
  • Killing aliens count as plus resources.

Aliens:

  • Resource income from built RT counts as plus res.
  • Gestating into a lifeform or building hive/rt/chamber counts as minus res.

 

It is clear that inspecting these values are not accurate how important the job the player did to win.

  • Good lerk can delay (spawncamp) hive losing against non HA/JP marines. (but not spends/earns res for doing that)
  • SG-ing/biting an RT is critical for the aliens/marines economy in the long run. (no res spend/earn for killing RT)
  • Using PG is not measurable in resources.
  • (gameplay is very complex) etc...

 

The main goal is to track everything ingame, like in Supreme Commander/World of Warcraft/Left 4 Dead/Team Fortress 2/etc... Moar stats are better then less. Write an event oriented AMX plugin including weld request, ammo taken from armory, spore damage, succesful parasites, wasted bullets, scan reveal cloaked aliens count, disappearing (wasted) medpacks, everything. It is not easy. :) And all events graphically illustrated on a top-down map view, like Episode2 stats.

qin
286
Posts
30
Location
Finland
Joined
11 May 2005
8 October 2009 - 10:20 CEST
#24

I want a plugin to keep track of how often I fail at chemical's ceiling jump.

Edit: On second thought, welding-stats might actually help choose teamplayers in gathers.

Edit2: Ultimately, this thread will lead us into having a 3rd system integraded to ENSL site; NS Manager pro. NS2 will come but people can still have fun with the old stats and mechanics, all text based.

lump
81
Posts
499
Location
United Kingdom of Great Britain and Northern Ireland
Joined
7 May 2005
8 October 2009 - 14:19 CEST
#25
actually shit yeh, i wanna see a stat on 'medpacks recieved'... that'd make me kinda  horny.
jiriki
176
old people
Admins
Provider & Webmaster
Posts
490
Location
Oulu, Finland
Joined
1 May 2009
8 October 2009 - 15:36 CEST
#26

I think Fana's concern is a valid one, however I think stats would be useful atleast when picking players in a gather alteast, or recruiting new people.

Stats are useful when you understand the underlying factors. Actually calculating individual player rankings from a team game is not impossible, it just requires advanced mathematics. There are several importants factors you need to take into concern:

  • Own players and opponent's players. 
  • Different roles and lifeforms
I'd say other factors (such as players dropping out, bad pings, not playing seriously enough etc.) are also an issue, but not nearly as important as the two above. For the first one, there is Microsoft's Trueskill making it a non-issue, however combining that with the 2nd one is very problematic as you would need some a) True Math Skills to produce the formula in the first place b) way to calculate some quantitative factor for different roles c) analyze the results.
The good thing about analyzing past results is that you don't have to worry that much about opponent selection and make an implicit assumption of normally distributed match pairs.

Miglecz, the main goal is that you pay for all that! Yeah I know how to make AMX log welding requests (actually I would have find the right NS message for that) etc but I don't if I cba for NS1, and all of them add overhead both log and game-wise. If money is not an issue, I'm sure we can find an agreement. ;)

Get to the spaceship!
lump
81
Posts
499
Location
United Kingdom of Great Britain and Northern Ireland
Joined
7 May 2005
8 October 2009 - 15:50 CEST
#27

I don't agree with any of that, being a good player is only shown through the end results as that's what the aim is.. but if you're the best team player around and your team is shit, it'll make you look worse because you're assuming teamplay that doesn't happen.

 

E.g.  a good teamplay strat is to attack as a group as soon as an opportunity is open to save time and gain fast control, if you go alone assuming the rest of the team has seen the same opportunity, you can end up getting slaughtered and the people who missed an opportunity get a few easy frags covering you.

 

Stats simply don't show anything but how good people are at getting stats and really don't reflect skill imo.. (apart from very long games with lots of jps covering locations for some reason)

 

Seriously, how much people cost in medpacks would be really interesting to see if their score is actually benefitial other than being a destraction.

Miglecz++
1643
Posts
42
Location
Debrecen, Hungary
Joined
7 February 2008
8 October 2009 - 17:50 CEST
#28

hey J!

did u read Praying Mantis (he was the author) article about calculating effectiveness of units in Supreme Commander with different weights?

that article is no more sadly :( but he used lotof maths to calculate lotof thing, it was impressive, if u remember that we could use a lot of thing to calculate stats

old site url (it is unreachable now) http://supcomfaguide.com/

it is sad that i not saved all content

can we get the old content from google somehow??

update: OLOL i was smart before, i have found it on my hard drive :DDD

sublime
5
Posts
488
Location
United Kingdom of Great Britain and Northern Ireland
Joined
6 May 2005
8 October 2009 - 18:10 CEST
#29

This game isn't like any other RTS you would be hard pushed to create a model that took into account half of things it needs to. It's not about lifeform X fights lifeform Y, calculate damage they can do to each other, come to a conclusion.

Miglecz++
1643
Posts
42
Location
Debrecen, Hungary
Joined
7 February 2008
8 October 2009 - 18:51 CEST
#30

the math theory is what counts not game type

and the practically collected observations

'This game isn't like any other RTS' ofc

http://supcomfaguide.uw.hu/

New Reply