PhysicsJunky Posted May 30, 2010 Report Share Posted May 30, 2010 (edited) With the help of TBRaiders I've compiled what is probably the most extensive survey of alliances to date. Along with the normal statistics kept by Unspeakable Evil these focus on the number of nations alliances have in the top 1%,5%,10%,25% & 50% of the total population. Sadly due to some holdups the date all of these statistics were taken was two weeks ago so there of course have been changes, disbandments and rebuilding efforts since. Additionally while we sampled 211 alliances some of the smallest alliances in CN were inevitably left off. I'd like to call special attention to the statistical significance categories I've included. Using probabilistic methods this ranks the eliteness of alliances with consideration of size, as opposed to the straight percentages which favor smaller alliances such as mine. With 60 pages worth of information I've had to make the file available as a small PDF rather than posting the numbers in thread. I apologize for the inconvenience. [url="http://www.mediafire.com/file/nzyjlwjztov/CN%20Census.pdf"]CN Census PDF[/url] [url="http://www.filedropper.com/cncensus"](If the first link is down)[/url] Edited May 31, 2010 by PhysicsJunky Quote Link to comment Share on other sites More sharing options...
Believland Posted May 30, 2010 Report Share Posted May 30, 2010 This is pretty cool. Way to go Quote Link to comment Share on other sites More sharing options...
Uralica Posted May 30, 2010 Report Share Posted May 30, 2010 Well colour me impressed. A most excellent job. Quote Link to comment Share on other sites More sharing options...
Biazt Posted May 30, 2010 Report Share Posted May 30, 2010 Quite impressive. You say these stats were collected 2 weeks ago? Apparently GOONS gained 58 members in that timeframe Quote Link to comment Share on other sites More sharing options...
dy Cazaril Posted May 30, 2010 Report Share Posted May 30, 2010 Having just taken Statistics, I find the concept of "statistical significance" for alliance strength distributions to be an interesting concept. Any chance you could elaborate on the methods you used to determine the significances? Quote Link to comment Share on other sites More sharing options...
KagetheSecond Posted May 30, 2010 Report Share Posted May 30, 2010 That was delightful. Thank you. Quote Link to comment Share on other sites More sharing options...
Banksy Posted May 30, 2010 Report Share Posted May 30, 2010 (edited) I see ODN Applicant continues to have an outrageously large number of nations. [quote name='Biazt' date='31 May 2010 - 09:26 AM' timestamp='1275251149' post='2316976'] Quite impressive. You say these stats were collected 2 weeks ago? Apparently GOONS gained 58 members in that timeframe [/quote] Yes- I noticed GOONS, UPN, GATO, NpO and the ODN were all short changed Edited May 30, 2010 by Banksy Quote Link to comment Share on other sites More sharing options...
Vilien Posted May 30, 2010 Report Share Posted May 30, 2010 Absolutely astounding collection of data and analysis. Truly incredible work. Quote Link to comment Share on other sites More sharing options...
PhysicsJunky Posted May 30, 2010 Author Report Share Posted May 30, 2010 [quote name='dy Cazaril' date='30 May 2010 - 04:27 PM' timestamp='1275251228' post='2316979'] Having just taken Statistics, I find the concept of "statistical significance" for alliance strength distributions to be an interesting concept. Any chance you could elaborate on the methods you used to determine the significances? [/quote] Binomial distribution off the sampled population, I also ran cumulative binomials which tell you the chance that if you were to randomly create an alliance with the same number of members you could improve. The cumulative mode is interesting but less meaningful in terms of ranking. Quote Link to comment Share on other sites More sharing options...
Zoomzoomzoom Posted May 30, 2010 Report Share Posted May 30, 2010 Poison Clan definitely needs to get its nuke count up. Our ratio is awful. If you ever do this again you should include land statistics. Pretty cool otherwise. Quote Link to comment Share on other sites More sharing options...
dy Cazaril Posted May 30, 2010 Report Share Posted May 30, 2010 [quote name='PhysicsJunky' date='30 May 2010 - 03:33 PM' timestamp='1275251611' post='2316989'] Binomial distribution off the sampled population, I also ran cumulative binomials which tell you the chance that if you were to randomly create an alliance with the same number of members you could improve. The cumulative mode is interesting but less meaningful in terms of ranking. [/quote] Finding binomial probabilities isn't the same as finding statistical significance, though. Significance would imply some sort of test using a random sampling, not finding a distribution based on a census of the entire population. Quote Link to comment Share on other sites More sharing options...
Lord Boris Posted May 30, 2010 Report Share Posted May 30, 2010 pheonomenal work. I must say, I'm quite pleased with how my alliance stacks up in a lot of them, considering our relative newness and minimal member base. Quote Link to comment Share on other sites More sharing options...
lonewolfe2015 Posted May 30, 2010 Report Share Posted May 30, 2010 This is amazing, kudos to TB and PJ, you guys did great work here. If you need a packhorse to gather information next time just ask and I'll give it a run. Perhaps we can turn this into something done every month or so with purely the statistical breakdown and without the graphs to reduce workload? Quote Link to comment Share on other sites More sharing options...
PhysicsJunky Posted May 30, 2010 Author Report Share Posted May 30, 2010 (edited) [quote name='dy Cazaril' date='30 May 2010 - 05:08 PM' timestamp='1275253698' post='2317009'] Finding binomial probabilities isn't the same as finding statistical significance, though. Significance would imply some sort of test using a random sampling, not finding a distribution based on a census of the entire population. [/quote] A binomial distribution tells you the probability of having exactly x number of some type of nations in an alliance of y nations, given p probability of a nation being of that type. It's a perfectly fine metric to use when ranking alliances, it is however a misnomer on my part to call it the "statistical significance" proper. It is A statistical significance, it's just not THE statistical significance. I didn't want to write a paragraph for each title, and the binomial is the type of probability your average CN player is likely more comfortable with. It's the chance an alliance of a certain size would have that many members in that percentile. I was more concerned with getting the rankings right. My bad. This took a good chunk of time. [quote]Yes- I noticed GOONS, UPN, GATO, NpO and the ODN were all short changed [/quote] No matter when you take stats in this game somebody is going to be under-represented. In this round NPO got the rawest deal. If I had done the footwork I'd have waited until terms ended but the raw data came to me as a gift and I wasn't going to complain. Edited May 30, 2010 by PhysicsJunky Quote Link to comment Share on other sites More sharing options...
Schad Posted May 30, 2010 Report Share Posted May 30, 2010 That's fantastic...great work putting this together. Quote Link to comment Share on other sites More sharing options...
Brutilius Posted May 30, 2010 Report Share Posted May 30, 2010 This is fantastic! Well done. Quote Link to comment Share on other sites More sharing options...
ratonbox Posted May 30, 2010 Report Share Posted May 30, 2010 Wow, this must have been a lot of work. Well done! Quote Link to comment Share on other sites More sharing options...
dy Cazaril Posted May 30, 2010 Report Share Posted May 30, 2010 [quote name='PhysicsJunky' date='30 May 2010 - 06:08 PM' timestamp='1275260863' post='2317133'] A binomial distribution tells you the probability of having exactly x number of some type of nations in an alliance of y nations, given p probability of a nation being of that type. It's a perfectly fine metric to use when ranking alliances, it is however a misnomer on my part to call it the "statistical significance" proper. It is A statistical significance, it's just not THE statistical significance. I didn't want to write a paragraph for each title, and the binomial is the type of probability your average CN player is likely more comfortable with. It's the chance an alliance of a certain size would have that many members in that percentile. I was more concerned with getting the rankings right. My bad. This took a good chunk of time. [/quote] Regardless of my labeling quibble, excellent job. It's good to see that GPA is doing well in the rankings. Quote Link to comment Share on other sites More sharing options...
Bob Ilyani Posted May 30, 2010 Report Share Posted May 30, 2010 [font=Arial, Verdana, Tahoma, sans-serif]Wow, this is a pretty incredible collection of stats. Will you be updating it frequently?[/font] Quote Link to comment Share on other sites More sharing options...
PhysicsJunky Posted May 30, 2010 Author Report Share Posted May 30, 2010 [quote name='Bob Ilyani' date='30 May 2010 - 07:40 PM' timestamp='1275262812' post='2317168'] [font=Arial, Verdana, Tahoma, sans-serif]Wow, this is a pretty incredible collection of stats. Will you be updating it frequently?[/font] [/quote] I think if this continues I'll definitely scrap the graphs sections to streamline it a bit. If there are volunteers to grab the raw statistics (the number of nations in each percentile needs to be done manually rather than grabbed from Unspeakable Evil) I could see this being an every two-three month sort of thing. I view its value largely in the community being able to look back over a long period of time and having a snapshot of what the CN world was like. Quote Link to comment Share on other sites More sharing options...
Groucho Marx Posted May 31, 2010 Report Share Posted May 31, 2010 Impressive work. Quote Link to comment Share on other sites More sharing options...
Bob Ilyani Posted May 31, 2010 Report Share Posted May 31, 2010 [quote name='PhysicsJunky' date='30 May 2010 - 07:52 PM' timestamp='1275263504' post='2317177'] I think if this continues I'll definitely scrap the graphs sections to streamline it a bit. If there are volunteers to grab the raw statistics (the number of nations in each percentile needs to be done manually rather than grabbed from Unspeakable Evil) I could see this being an every two-three month sort of thing. I view its value largely in the community being able to look back over a long period of time and having a snapshot of what the CN world was like. [/quote] Sounds great. Will definitely be using this. Quote Link to comment Share on other sites More sharing options...
commander thrawn Posted May 31, 2010 Report Share Posted May 31, 2010 Asgaard doing it right Quote Link to comment Share on other sites More sharing options...
Prodigal Moon Posted May 31, 2010 Report Share Posted May 31, 2010 I can't wait to check this out but the link seems to be down. Quote Link to comment Share on other sites More sharing options...
Haflinger Posted May 31, 2010 Report Share Posted May 31, 2010 Nice one. Will be interesting to compare especially NPO's stats if this gets updated in a year's time or so (assuming they're not under terms again, heh). Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.