Online Player Statistics

[kaen]

As many of you are aware, Bitfighter has been collecting detailed stats on players and games for quite some time. This weekend I put up the first version of a player statistics page where you can view this information on a per-player basis.

For the impatient, go here: http://bitfighter.org/stats/stats.php

These stats are sent to the master server at the end of each game, and this page is updated immediately when that happens. Currently stats reflect the totals for all of time, but ways of segmenting the data (perhaps by month) are planned.

You can click on player names in this list to see a more detailed breakdown of the stats as well as per-weapon accuracy and firing statistics. You can query the list by player name (it will show all partial matches) as well as sort it by any column.

Also, don't forget about the game reports that are (and have been) available at http://www.bitfighter.org/stats/gamereports.php

Your feedback is very much welcomed.

Edit (6/30/2012):
Updated today. New features:

• Stats by month (or all-time)
• Separate entries for authed/unauthed players
• Authenticated players are underlined and bolded
• Total games played column
• "Advanced search" feature (only one "advanced" option right now: filtering on authentication status)
• Switching between authed/unauthed stats in individual player stats

Some known issues and Potentially Frequently Asked Questions:

Why are there authenticated entries for "yourname.0"?
These are instances where the player ("yourname") has logged in to a server twice while authenticated, causing his name to be appended with ".0"

However, this has been fixed in upcoming version. Now joining a server twice while authenticated will kick your old connection.

It looks exactly like it did before!
It definitely looks similar, but if you saw the last version you probably just need to clear your browser's cache. There should be noticeable differences. If you haven't visited the stats page before the last update, clearing your cache is not necessary.

Edit (12/26/2012):
Bitfighter Stats now has its own google code repository. Bugs can be reported at https://code.google.com/p/bfstats/issues/entry where they can be addressed in an organized and public manner.

Edit (2/22/2013):
Badges are now displayed on the individual player stats pages.

Edit (3/3/2013):
The following new stats have been added to the individual player stats pages:

• Asteroids Destroyed
• Force Fields Destroyed
• Turrets Destroyed
• Force Fields Engineered
• Turrets Engineered
• Teleporters Engineered
• Distance Traveled

Please note that everybody's totals for these columns started at zero just a few days ago.

[raptor]

This is great!

[FoOtloOse]

YAY!

[Skybax]

I love my Mine hit percentage

[ZoombeR]

This is very neat, something I've always wanted to see.

[Santiago ZAP]

I see playing fast nexus improved my death percentage xD
I love the maker of that!

This is great. :3
Maybe I should stop dying and start killing. Anyway, people sure do have strange names.

Nice work! I especially love the weapon statistics - apparently I've fired phaser for over 6.5 hours! I have some things to say if you don't mind feedback:

» Does Bitfighter track IP addresses or any way to monitor for players with changing usernames? "3296 players in database" is a bit misleading of our player count, and I'd be curious to see what our actual numbers are anyway. Along the same lines, if someone lags out and rejoins under the ".0" name, it would be helpful to know that they are the same player and then merge those statistics.

» A "total games" column under the Games parent column should be added if possible.

» Would it be feasible to implement an option that hides players who haven't been online within x number of months? Sorting by KDR, the top six players haven't been on for at least one year, which doesn't give me a very accurate idea of the active hotshot players. However, hiding all players who have been online in the past 3 days might give me a better idea of who's good and who's not.

» Another filtering idea would be to hide players who haven't participated in less than x games. You'll notice that the active players don't have higher than a 1.5-2.0 KDR, so ordering by max KDR to determine the best players is unhelpful (when sammysss has 57.0 KDR).

» I might also suggest using the "players online" feature from the forums and putting it somewhere at the top of the spreadsheet. No real reason except for completeness and me being picky.

» Finally, does your list do anything other than sort & prettify the incoming data? I saw a couple of bugs (players being online within minutes not showing up on the last seen column, _k playing 1500+ games but not stealing a single flag, etc). I'm curious as to which end is the buggy end.

Thanks! Great job!

EDIT: Sorry to overwhelm ya, but I thought of something else - can Bitfighter track player time, in the context of "Joe played for 27 minutes starting at 18:04 UTC". I would think it can since it's recorded in the game reports but I wanted to confirm. Something that might be helpful would be to compile and aggregate that data over a period of a month or so, to determine when players are most active, and predict when the best gaming times are. Just a thought!

[kaen]

Hey, nice reply!

I have some things to say if you don't mind feedback:

I would liken myself to an oscillator, which is to say that feedback is a requirement for me to meaningfully function.

» Does Bitfighter track IP addresses or any way to monitor for players with changing usernames?

I have not seen any IP addresses in any of the databases, so I expect not. It's possible, we could even just use a hashed value. IP addresses aren't permanent or truly unique to one person (usually), so the benefits are marginal in my opinion. However, there is auth data present. Applying this is high priority for me.

» A "total games" column under the Games parent column should be added if possible.

Will do.

» Would it be feasible to implement an option that hides players who haven't been online within x number of months?

My tentative plan is to have data partitioned by month, and kept for six months (or so). So every month stats reset to 0, but you can view stats for recent months.

» Another filtering idea would be to hide players who haven't participated in less than x games

» I might also suggest using the "players online" feature from the forums

I like both of these .

» Finally, does your list do anything other than sort & prettify the incoming data?...

Honestly I've been puzzling over this all night. I have a sinking sensation that I've malformed a query (or trigger...), but let me know if you find any other instances of that sort of thing!

Something that might be helpful would be to compile and aggregate that data over a period of a month or so, to determine when players are most active...

I don't think that specific information on how long each player has played is kept, only the time that the data was received by the master server. The active play time projection sounds feasible with current data, however.

Thanks all for the kind words, suggestions, and especially bug reports .

[sam686]

I saw a couple of bugs (players being online within minutes not showing up on the last seen column, _k playing 1500+ games but not stealing a single flag, etc). I'm curious as to which end is the buggy end.

The problem is, players stats isn't logged or added up, and update last seen until the game ends, that is if player stayed in game.

Teleportations, Asteroid Crashes, Flags taken, dropped, return, scored, was not logged in Bitfighter version 017, as those stats are zero on Bitfighter 016 and earlier. Players who never played Bitfighter 017 will make those stay at zero.

As for stats, I like to see separate stats between Authenticated and non-Authenticated players, just like having separate stats between different player names.

[Lone Wolf]

This is great. :3
Maybe I should stop dying and start killing. Anyway, people sure do have strange names.

You only have 1 death and no kills. xD

[kaen]

Edited original post.

[sam686]

As for the new player stats:
1. Why is there is stats for July 2012 when it is not July yet? The old game reports have nothing for July 2012, yet. Also noticed some of "Last seen" date is in the future. Time zone?
2. When going back a month and clicking several player name, it goes back to the latest month and says "Name not found".

[kaen]

1. Why is there is stats for July 2012 when it is not July yet?

It's been July for several hours where the server is located (GMT + 4). I haven't adjusted the time stamps because they're a stop-gap until I can make the time-elapsed function work in all the browsers (have to do it client-side because of caching, otherwise it'll almost always be wrong).

2. When going back a month and clicking several player name, it goes back to the latest month and says "Name not found".

The page didn't remember the time period you were querying for. I just fixed that. Let me know if I misunderstood you.

Thanks!

@sam Thanks for clearing up the issue with some stats not being recorded.

@kaen Thanks for reading and responding quickly to my ideas! New format looks fantastic, but I'm not seeing the bold/underline for registered players...?

Also, maybe it's just me being picky but for aesthetic reasons I think combining the filter-by-month links into a drop-down menu might look a bit nicer, as right now it's crammed. A drop-down menu also ensures that as more and more months are added, the line of text won't get crowded.

EDIT: nevermind about not seeing bold & underline, cache must've had an out-of-date version of the spreadsheet. that also cleared up my concern about aesthetic appeal with the months -- erm, it looks much better now that I'm actually seeing the up-to-date version (FML) so disregard that request.

[Skybax]

I am..

SO cool! Lol I never shot or got shot by triple

[kaen]

Bump for the new update.

suggestions.

1 time cloaked

2 time shielded

3 time boosting

4 number of speed zones used

5 nubmer of time blasted by turret

6 forcefeilds broken.

7 time idle

8 nubmer of objects engenerred

9 asteroids destoryed

10 longest time played

11 average time played

12 time in editor

13 levels made

[kaen]

Noted, but none of those things are currently tracked, amgine, so the summed likelihood of any of those being added is <1%

its ok if you dont do it i just think it would be intresting.

also 14 . shots fired.

15 . accuracy

[kaen]

Those two are both already reported:
http://bitfighter.org/stats/player.php? ... 13&month=2

16 distance traveled.

17 server join count?

jsut some ideas.

[watusimoto]

Distance traveled is a nice one -- completely meaningless, and yet... somehow interesting.

A lot of this can be tracked by the server, and could thus be added even during a minor update. I'll take a look.

I agree, knowing how many miles my ship has traveled in its lifetime would be completely pointless yet totally intriguing.

@kaen stats page looks way sexy.

@lamp how is your teleportation count so HIGH!? What are you doing!? It's at 15,570 for all time, but the real kicker is that 12,370 (80%) of those are in February alone. Which means that your average for this month is close to 500 teleportations per DAY.

[bobdaduck]

kaen: Bitfighter stats should probably not count bobdaduck.0 as a seperate user. There's a lot of random "fake" stats floating around... People who played two games and have 50 KDR for all time.

[kaen]

@kaen stats page looks way sexy.

Wait until you see the next update...

kaen: Bitfighter stats should probably not count bobdaduck.0 as a seperate user. There's a lot of random "fake" stats floating around... People who played two games and have 50 KDR for all time.

It definitely shouldn't, but that's a product of how stats are sent to the master (purely by name and authentication status). The real solution to the first part is to make it so servers kick old logins when an authenticated user rejoins rather than just adding a ".0" etc onto the name.

For the second point, I was against doing all-time stats at all (it essentially doubled the amount of code I had to write). Plus, as you mentioned, they don't decay so any ChumpChange can open up four bitfighter instances, set a server password, and become THE BEST BITFIGHTER OF ALL TIME by zerging. (Stats are really insecure, you could actually do much worse things with less work than I described, but that's an exercise for the reader).

I've been trying to think of a better default column to sort on, KDR is obviously a terrible choice. I was thinking either Last Seen, Games Won, or Kills. I welcome any input on this matter.

[bobdaduck]

@kaen stats page looks way sexy.

Wait until you see the next update...

kaen: Bitfighter stats should probably not count bobdaduck.0 as a seperate user. There's a lot of random "fake" stats floating around... People who played two games and have 50 KDR for all time.

It definitely shouldn't, but that's a product of how stats are sent to the master (purely by name and authentication status). The real solution to the first part is to make it so servers kick old logins when an authenticated user rejoins rather than just adding a ".0" etc onto the name.

For the second point, I was against doing all-time stats at all (it essentially doubled the amount of code I had to write). Plus, as you mentioned, they don't decay so any ChumpChange can open up four bitfighter instances, set a server password, and become THE BEST BITFIGHTER OF ALL TIME by zerging. (Stats are really insecure, you could actually do much worse things with less work than I described, but that's an exercise for the reader).

I've been trying to think of a better default column to sort on, KDR is obviously a terrible choice. I was thinking either Last Seen, Games Won, or Kills. I welcome any input on this matter.

Win/loss ratio.

[kaen]

Simple ratios are a bad choice in general because of the difference in sample sizes. A slightly better scheme is usually to factor in the sample size. For instance, instead of

  Code:

W / L

one could use

  Code:

(W / L) * (W + L)

although this can still act erratically for small values of (W + L).

Really, it should be weighted so that ties affect the value correctly. I'm not sure how to do that off the top of my head, but obviously ties are better than losses and worse than wins.

YAY NEW STATS PAGE :D

[watusimoto]

There's a lot of random "fake" stats floating around... People who played two games and have 50 KDR for all time.

These are legit, though possibly misleading when taken out of context. Like all the players with a 1000 batting average littering the Baseball Encyclopedia, who were only ever up once.

[watusimoto]

The real solution to the first part is to make it so servers kick old logins when an authenticated user rejoins rather than just adding a ".0" etc onto the name.

This is very difficult as the server does not yet know that kaen has been disconnected when kaen.0 rejoins. A better solution might be to auto-rename kaen.0 to kaen when kaen disconnects, and disallow users from choosing a .0 name. Or come up with some other custom name extender than can only be assigned by the server.

Just as a note of trivia, I once spent an evening figuring out how to make the names work if a player had the maximum length name, and then joined over 100 times simultaneously... there's special code in there to handle that.

[watusimoto]

I'm not sure how to do that off the top of my head...

Like this:
http://en.wikipedia.org/wiki/Elo_rating_system

This system can be extended to multiple player games. It's complicated, but we have the data to do it.

TRIPPLE POST C-C-C-COOMMBBOO BBRREEAAKKEERRRRR

[kaen]

In the latest mercurial, stale logins are kicked when an authed player rejoins. My servers are now running this code.

[kaen]

New stats added. See first post.

hooray how about......

1 average speed.
2. speed pads touched.
3. time in editor
4. curves use count in editor
5 . number of levels added to servers.
6 . most players on your server.
7 . average players on your server.
8. longest time a player was on your server.
9. longest time your server was up.
10. times you've kicked someone on your server.
11. times you've banned someone on your server.
12 shortest ban on your server.
13. longest ban on your server.
14. average ban on your server.
15 times you've been kicked.
16 times you've been banned
17. longest ban you've had been given.
18 shortest ban you have been given.
19 average ban you have been given.
20. number of different servers joined.
21. most joined server.
22. least joined server.
23. longest time on a server.
24. shortest time on a server.
25. average time on a server
26. time your server has been suspended
27. time you've been in a /suspend
28. average suspended server time.
29. times you've used /shuffle
30. times you've been shuffled.
31. times zapped by a bot.
32. bots added
33. times you've zapped bots =)
34. angle distance turned clockwise.
35. angle distance turned counter clockwise
36. total timed running BF
37. time spent in lobby
38. number of times changing modules
39. most used module.
40. least used module.
41. most used weapon
42. least used wepon.
43. global chat count.
44. team chat count.
45. voice chat use count.
46. longest continues voice chat usage.
47. longest chat message.
48. longest time not chatting.
49. credits watch count.
50. times on leader board
51. leader baord veiw count.
52. mode play counts ( rabbit soccer ect.)
53. most played mode
54. least played mode.
55. BBB participation count
56. level design contest join count. (level submtion count.)
57. times gone idle
58. longest time idle
59. average time idle
60. most kills before being killed
61. most times killed before getting a kill
62. who've you most killed.
63 who've you been killed most by.
64. times logged in. (registered.)
65. times logged out. (registered.)
66 times logged in with a non registered user name.
67. most flags put in nexus
68 most points from nexus drop.
69. average. nexus points score
70. average flags in nexus.
71 longest time with carrot.
77 average time with carrot.
78 damage dealt. (total of all.)
79 damage taken. (Total of all.)
80 cores zapped
81. core damage (given.)
82 force field damage (given.)
82. turret damage. (given)
83. turret damage. (taken.)
84. asteroid segments blasted.
85. asteroid damage given ( 1 point per segment.)
86. asteroid damage taken.
87. bot damage (given.)
88. bot damage (taken.)
89. opponent ship damage (given.)
90. opponent ship damage (taken.)
91 personal damage taken (total.)
92. distance traveled on x axis.
93. distance traveled on y axis.
94 x/y distance travel ratio.




( note 78 -91 ignoring 80 are x6 sorted by weapon.)


I challenge the admins to add all of these in a 1 year period


sorry for the over post.....

95. social security number
96. creditcard number
97. photo ID
98. hight and weight
99. date of birth

Haha Blackbird.

Amgine, I am all for big brother-esque statistic tracking, but could you maybe prioritize that list? Are there 2-3 features that you'd really love to see tracked more than any others?

[watusimoto]

I shortened the list to those that I think are conceivably of interest. Those marked with a ==> are ones I think we could extract from our currently collected stats, if we really cared.

I've added one or two to my todo list.

2. speed pads touched.
==> 6 . most players on your server.
==> 7 . average players on your server.
==> 8. longest time a player was on your server.
10. times you've kicked someone on your server.
11. times you've banned someone on your server.
15 times you've been kicked.
16 times you've been banned
==> 20. number of different servers joined.
==> 21. most joined server.
==> 22. least joined server.
34. angle distance turned clockwise.
35. angle distance turned counter clockwise
36. total timed running BF
==> 39. most used module.
==> 40. least used module.
==> 41. most used weapon
==> 42. least used wepon.
==> 50. times on leader board
==> 52. mode play counts ( rabbit soccer ect.)
==> 53. most played mode
==> 54. least played mode.
60. most kills before being killed <== streak?
61. most times killed before getting a kill <== anti-streak?
62. who've you most killed.
63 who've you been killed most by.
67. most flags put in nexus
==> 68 most points from nexus drop. (if we have 67)
71 longest time with carrot.
77 average time with carrot.
78 damage dealt. (total of all.)
79 damage taken. (Total of all.)
80 cores zapped

3 31 33 36 37 and 43 would be nice.

your correct on 60 and 61.

also time spent /idle (total) and average idle time. (different from being on a suspended server.)

but yes i agree those you picked would a be a nice priority. the other meh.

"Ive added 1 or 2 to my to do list" : lol watssimo t6hats like 20 things. XD

I haven't played this game for a year and somehow I'm 10th on the all-time kills list. #skillz

[kaen]

Well if we're gonna brag, I'm the 12th rated flag returner for March 2013

> get on my level

Tempting, but as #1 by a margin of 150+ returns, I'm going to pass.

amgine once again leading the "teleportation" area of the scoreboard with more than 100,000 teleports over the course of a month. Seriously, wtf?

[Quartz]

Well if we're gonna brag, I'm the 12th rated flag returner for March 2013

> get on my level

rofl

amgine once again leading the "teleportation" area of the scoreboard with more than 100,000 teleports over the course of a month. Seriously, wtf?

...I need to beat him at this.

tele spam level XD

i believe i also hold the distance record......

[Santiago ZAP]

Bro, I hold the more nexus points scored in the whole universe!
(*cough* fast nexus *cough*)

[kaen]

Version 2.0 released! See viewtopic.php?f=20&t=2239&p=22536#p22536