These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

EVE General Discussion

 
  • Topic is locked indefinitely.
Previous page123Next page
 

Analyzed the connections between systems using network metrics and posted a visual- You'll Love

First post
Author
Joe Themachine
Sleeper Tech. Research Foundation
#21 - 2013-12-12 22:03:40 UTC
Pew Terror wrote:
Joe Themachine wrote:
Pew Terror wrote:
Running specialized metrics for specific use cases (in your case network analysis) on generic graphs is as non scientific as it gets.
You need to demonstrate what analysis the tool you used does and then present a case why the asssumptions of the model concerning its nodes and edges holds in this specific case.
You can't just go "Im a gonna force this data through this tool i just learned existed because pretty color, look at me!".


Who said anything about scientific? It was mostly fun, and I was testing my code...:) you don't see any tests of significance do you? thanks for your insight..


Testing for significance is part of statistics, not graph theory (what your tool there did where you clicked that button) and it a completly different area of mathematics.
Sticking a finger in your bum can be fun as well, but the world dont need to hear of it when its not useful information.


Look up the QAP permutation model which is also used by UCINET... testing for significance is an ESSENTIAL part of network modelling especially for stochastic methods...
Mara Rinn
Cosmic Goo Convertor
#22 - 2013-12-12 22:05:30 UTC
Joe Themachine wrote:
******(if you search for Jita or Rens, you'll see that they are neither, which means that they are unnatural trade hubs--all marketing and word of mouth--if the forums didn't exist Jita and Rens WOULD NOT be trade hubs) *********


Entirely missing the nature of systems in EVE (they are not all equal, a system with a high number in your model will not necessarily have high traffic or connectivity).

Entirely missing the reason trade hubs become trade hubs (the network effect).

You can't derive any conclusions about "natural hubs" from your data or your pretty clicky web toy. What you can derive is that when you mash Data X through Process Y you get Result Z. Result Z might be pretty according to some people, but it is meaningless.

If you want to derive information about "natural hubs" you would need to demonstrate what you mean by that term, then produce some models to illustrate how your technique can locate these "natural hubs" and then apply that technique to the network of interest.

Relevant data that you haven't accounted for:

  • Mission hubs (easy to find, look for NPC kills in last 24 hours)
  • Production hubs (harder to find, look for volume of ore mined in last 24 hours, there are proxy values available)
  • Industry hubs (look for S&I jobs completed in last 24h)


And then there's the network effect which you can't account for: everyone goes to Jita because everyone goes to Jita. That's as natural and organic a trade hub as you are ever going to find.
Miilla
Hulkageddon Orphanage
#23 - 2013-12-12 22:06:52 UTC
Joe Themachine wrote:
Pew Terror wrote:
Joe Themachine wrote:
Pew Terror wrote:
Running specialized metrics for specific use cases (in your case network analysis) on generic graphs is as non scientific as it gets.
You need to demonstrate what analysis the tool you used does and then present a case why the asssumptions of the model concerning its nodes and edges holds in this specific case.
You can't just go "Im a gonna force this data through this tool i just learned existed because pretty color, look at me!".


Who said anything about scientific? It was mostly fun, and I was testing my code...:) you don't see any tests of significance do you? thanks for your insight..


Testing for significance is part of statistics, not graph theory (what your tool there did where you clicked that button) and it a completly different area of mathematics.
Sticking a finger in your bum can be fun as well, but the world dont need to hear of it when its not useful information.


Look up the QAP permutation model which is also used by UCINET... testing for significance is an ESSENTIAL part of network modelling especially for stochastic methods...


Your underlaying STRUCTURE of the Eve Universe "network" is INCORRECTLY MODELLED, your model does not reflect the system under analysis.
Ambassador Crane
Hellhound Productions
#24 - 2013-12-12 22:23:11 UTC
Posting in a thread I have absolutely no idea what the OP was saying and even less so what these 3 are arguing about. But I can say it sounds like Miilla is having a bad day or something.

On the other hand, OP sounds like he's having fun. (though how any of that hubla dubla wada wada could be fun i dunno Ugh but more power to ya )

All I can say is, it looks pretty!
Joe Themachine
Sleeper Tech. Research Foundation
#25 - 2013-12-12 22:26:10 UTC
Ambassador Crane wrote:
Posting in a thread I have absolutely no idea what the OP was saying and even less so what these 3 are arguing about. But I can say it sounds like Miilla is having a bad day or something.

On the other hand, OP sounds like he's having fun. (though how any of that hubla dubla wada wada could be fun i dunno Ugh but more power to ya )

All I can say is, it looks pretty!


LOL! :)

Magna Mortem
The Scope
Gallente Federation
#26 - 2013-12-12 22:33:39 UTC  |  Edited by: Magna Mortem
Joe Themachine wrote:
Ambassador Crane wrote:
Posting in a thread I have absolutely no idea what the OP was saying and even less so what these 3 are arguing about. But I can say it sounds like Miilla is having a bad day or something.

On the other hand, OP sounds like he's having fun. (though how any of that hubla dubla wada wada could be fun i dunno Ugh but more power to ya )

All I can say is, it looks pretty!


LOL! :)

Why did you, right from the start, ignore the possibility that she might be right and you might be wrong?

Your response being "look up that book at page 2" is completely pointless. Most people in here, including myself, have no clue about what you are doing and we absolutely deserve that it is explained to us in a proper manner, considering that you want us to take you seriously. And that's what you want, else you wouldn't have posted.

Of course I do recognize that another reason is that you wanted attention, but I do not care about that.


So, instead of giving others reasons to make fun of you, which is then leading to you seriously giving them forevermore reason to actually rightfully make fun of you, prove that you can debunk their claims and actually
know what you are talking about.



Thank you very much.
Joe Themachine
Sleeper Tech. Research Foundation
#27 - 2013-12-12 22:59:57 UTC  |  Edited by: Joe Themachine
Magna Mortem wrote:
Joe Themachine wrote:
Ambassador Crane wrote:
Posting in a thread I have absolutely no idea what the OP was saying and even less so what these 3 are arguing about. But I can say it sounds like Miilla is having a bad day or something.

On the other hand, OP sounds like he's having fun. (though how any of that hubla dubla wada wada could be fun i dunno Ugh but more power to ya )

All I can say is, it looks pretty!


LOL! :)

Why did you, right from the start, ignore the possibility that she might be right and you might be wrong?

Your response being "look up that book at page 2" is completely pointless. Most people in here, including myself,
have no clue about what you are doing and we absolutely deserve that it is explained to us in a proper manner,
considering that you want us to take you seriously. And that's what you want, else you wouldn't have posted.

Of course I do recognize that another reason is that you wanted attention, but I do not care about that.


So, instead of giving others reasons to make fun of you, which is then leading to you seriously giving them
forevermore reason to actually rightfully make fun of you, prove that you can debunk their claims and actually
know what you are talking about.



Thank you very much.


No one really KNOWS what they're talking about when it comes to network science, including me, but that's not the point. Magna... I already acknowledged that the graph has two edges on each system when it only should be one (I think the second or third post), but because the analysis is structural and not statistical, centrality and network measures yield the same conclusions...they are based on the position of the planetary system, relative to other systems..just because the graph has two links instead of one for each jump does not change the analysis...

think of it as A = b x C where b is a scalar and C is a vector. The result, A, in this case, continues to be a vector regardless of the size or sign of b.

I already said the analysis is the same so I'm not sure what else you want... if you don't like my analysis, you are always welcome to post your own...thanks for your contribution.
Preto Black
Solar Clipper Trading Company
#28 - 2013-12-13 02:26:36 UTC

I admire the attempt to provide some investigation of the structure of the Eve universe and link it to matters like gate camping and appropriate trade hubs.

However this analysis does not seem to provide much for Trade Hubs.

Oursalaert III is the largest of the 2nd rank trade hubs (ie 6th overall) yet has a relatively small EVC value and on the face of it ought to be a classic candidate for your proposition given that networking and word of mouth drag most off to the largest hubs, theoretically leaving more structural effects to be shown in the 2nd rank hubs.

The 7th largest hub VFK-IV has a tiny EVC, but presumably also suffers from popularity due popularity's sake.
Pinky Hops
Caldari Provisions
Caldari State
#29 - 2013-12-13 02:51:52 UTC
Good job op at attempting to unravel the graph network a bit. It will take a lot more work to extract new and useful things, though.

Ignore the obvious troll in the thread. Just because a gate doesn't have a specific "direction" associated with it does not in any way mean that displaying data in the form of a digraph is not appropriate. Trade and motion naturally has a flow, and thus is naturally better presented in the form of a digraph.

Keep it up, OP.
Daenika
Chambers of Shaolin
#30 - 2013-12-13 03:12:25 UTC
Am I the only one that noticed that Miilla spent like 3 pages posting damn near every other post, saying the exact same thing (it's not a directed graph!!!!), and no one listened?

Fun fact, Miilla: a directed graph where both directions are weighted equally is, by definition, an undirected graph.
Pinky Hops
Caldari Provisions
Caldari State
#31 - 2013-12-13 03:33:41 UTC
Daenika wrote:
Am I the only one that noticed that Miilla spent like 3 pages posting damn near every other post, saying the exact same thing (it's not a directed graph!!!!), and no one listened?


Lol

Daenika wrote:
Fun fact, Miilla: a directed graph where both directions are weighted equally is, by definition, an undirected graph.


I agree with your sentiment, but technically this isn't true as stated. They are isomorphic, but they are not exactly "equal" in any rough sense as one of the structures has extra ornaments attached, and the other does not.
Jessica Danikov
Network Danikov
#32 - 2013-12-13 06:26:12 UTC  |  Edited by: Jessica Danikov
I don't know what's crawled up Miilla's ass that made them what to troll the thread, but don't feed the troll I guess.

Daeniika's absolutely right that an undirected graph is a special subclass of digraphs (to be precise, it is a symmetric (all arcs have a matching return arc) unweighted (all arcs have equal weight) digraph).

Having the digraphs features available may yet turn out to be useful if you wish to use the asymmetric weighting. For example, you may wish to take in the average distance between stargates if you get that data available (especially considering the recent changes to warp speeds) or even map them all the warps and jumps individually (probably best to collect the distances and calculate time to warp for weighting based on the ship type).

It is most curious that the regions closely align with the regions CCP chose. Given that the recent blog on balancing the universe they disclosed that the constellation used to be the atom of node balancing and that a region is a collection of constellations, you're quite right that they probably did some similar static analysis (lacking any real data) when creating the universe.
Shaishi Otichoda
Science and Trade Institute
Caldari State
#33 - 2013-12-13 11:05:10 UTC
Quote:
Systems with high levels of EVC are great for trade. they are (probability-wise) are more likely to have items bough from, plus because high EVC systems are clustered together, they form clusters of trade hubs ******(if you search for Jita or Rens, you'll see that they are neither, which means that they are unnatural trade hubs--all marketing and word of mouth--if the forums didn't exist Jita and Rens WOULD NOT be trade hubs) *********
I think you may need to expand you definition of "natural" a bit. Jita is quite good location as it does have access to several different region just within two jumps. When you think about how EVE market works, this is quite important.

You want to price check several regions? There are three more just two jumps away.
Inzax
#34 - 2013-12-13 12:40:30 UTC  |  Edited by: Inzax
My eve graph.

probably should get out more.
Ambassador Crane
Hellhound Productions
#35 - 2013-12-13 13:06:49 UTC
Inzax wrote:
My eve graph.

probably should get out more.



Now THIS I can understand and support!
Joe Themachine
Sleeper Tech. Research Foundation
#36 - 2013-12-13 19:10:39 UTC
Inzax wrote:
My eve graph.

probably should get out more.


Hilarious!! You know if I had just thought ahead enough to just do that there would've been no complaining...:)
Nantwig Mutbrecht
Perkone
Caldari State
#37 - 2013-12-13 20:08:14 UTC
i would actually like a word of CCP explaining HOW the universe was ACTUALLY made. they could not only stop this thread, but also i would LOVE to know how the eve-universe was made.
that knowlege might even be usefull ingame. (say exploration fo rexample)
Josef Djugashvilis
#38 - 2013-12-13 21:33:27 UTC
I would love to pretend that I understand any of this thread, but in attempting to do so, I would only make myself look even more foolish than I normally do.

This is not a signature.

stoicfaux
#39 - 2013-12-13 21:45:46 UTC
Miilla wrote:
You have all graph vertices as in and out degrees, however, that is not how the Eve unvierse is, you are plotting one edge as two.

I fail to trust your intrepertation of the data as it seems incorrect. Eve universe is not a Di-graph.

Does that have any meaningful impact on the results and/or usefulness of the model? Meaning, is the OP's directed graph equivalent to a non-directed graph?

Pon Farr Memorial: once every 7 years, all the carebears in high-sec must PvP or they will be temp-banned.

Chopper Rollins
hahahlolspycorp
Brave Collective
#40 - 2013-12-13 22:03:11 UTC
As far as i can tell, OPs pretty picture measures system connectivity and provides no useful info other than that. Factors like the difference in sec status (that Rancer workaround hurts trade routes) and number/type of agents/stations/belts in each system all affect how systems are used.
Or not.
Such a pretty pattern would be thrown up from any complex map measured this way, telling us the most efficient WHATEVAH.

Humans on the other hand are not efficient, physically they are modified from an earlier form and full of terrible inefficiencies and redundancies. Optic nerve passing through retina? Pshaw!
Neurologically they are replete with spare backups, terrible inefficient if the human brain is seen as either a machine or as a computer. From an engineering perspective, it's a receiver.
Socially, humans are awful as far as efficiency goes, logic doesn't apply to a mess made up of units that can change roles to become parent to children that aren't theirs, provider to strangers and guardian to enemies as the need arises.

Systems theory used to be applied to the ecology, hoping to describe an underlying order that never existed. Ecological balance has been force fed to us and we still use it as a comforting description of the actual ocean of chaos we live in.

Factor in laziness, fear of combat, availability of services and friendship, then we'll have a more credible graph we can throw away.


Goggles. Making me look good. Making you look good.

Previous page123Next page