There can’t be any large-scale revolution until there’s a personal revolution, on an individual level. It’s got to happen inside first.-Jim Morrison
OK, so you guys have heard of the "six degrees of separation?"
Well, what follows is the degrees of separation from one member to another.
This is a huge table. If your computer can handle it, it tells you how many "hops" it takes to get from one member to another. Sometimes, it is impossible, and the table has "infinity" for that entry.
Table of Degrees of Separation
This is a histogram of the above data. Again, this is very typical of naturally occuring networks, a mode of 3 degrees of seperation is exhibited here.
---
Someone wanted a link to a full sized version of the picture I posted in th OP.
Again, this is a huge file. If your computer can handle it, here it is. The onlything more you may be able to get from this is to be able to make out the username. There are just way too many arrows.
--
I'll collect some of the post-count and join date data later tonight. Thus far I have focused only on the "network" aspects of the network.
Accept the past. Live for the present. Look forward to the future.
Robot Fusion
"As our island of knowledge grows, so does the shore of our ignorance." John Wheeler
"[A] scientist looking at nonscientific problems is just as dumb as the next guy." Richard Feynman
"[P]etabytes of [] data is not the same thing as understanding emergent mechanisms and structures." Jim Crutchfield
Can you feed the data in a self-organizing map, to see the what kinds of groups are "close" to each other and which are distant? Labels by type. I know it's much to ask if the data isn't readily useble for that kind of processing.
SOM is a really awesomely powerful tool, though!
I tried a bit to do it on my own. But I'll check out SOM. My Matlab version is a little newer than the one they listed for the package. Also, I have family comming to stay through Monday, so I may not get to it till next week.
Still, you can look ate the degrees of seperation to get an idea of how "close" two members are to each other.
Also note that most members are just three hops away from most other members, so I am not sure SOM would do a much better job than my initial attempt.
Sorry Elaur, I just didn't feel like looking back to see who it was while I was typing that post. Probably, not helping case here am I?
Rest assured, I usually notice it is you who write your posts. I just haven't had the greatest memory of late.
Accept the past. Live for the present. Look forward to the future.
Robot Fusion
"As our island of knowledge grows, so does the shore of our ignorance." John Wheeler
"[A] scientist looking at nonscientific problems is just as dumb as the next guy." Richard Feynman
"[P]etabytes of [] data is not the same thing as understanding emergent mechanisms and structures." Jim Crutchfield
SOM places similar items near each other, so persons having mostly similar friends would be placed near each other in the map with high probability, and persons with large distance between them would be placed far away in the map with high probability.
Or, that would be a proper goal when choosing the parameters for the data input.
I actually did group similar items together. The problem is everything is similar because other than a handful of items, everyone is connected together, and some members have 100+ connections.
I still haven't had a good chance to check out SOM.
It would be pretty easy to change the color to something else, but making it multi-colored would take some effort.
-----------------------------------
OK, so it's the weekend, and while I was hanging out on vent, I decided to take a few more stats.
So the correlation between # of friends and post count is 0.71
Here is the regression line and the data:
If we through out the 0 post count data (because log(0)=-infinity), and look at the correlation between the log of the # of friends and the log of the post count, we find that it is 0.78--a little better.
Here is the regression line and the data:
The correlation between # of friends and days on the forum is 0.23 A weaker correlation than between the # of friends and post count.
Here is the regression line and the data:
If we look at the log vs. log correlation, like we did with the post count, we find a stronger correlation, 0.30, but still fairly weak.
Here is the regression line and the data:
Accept the past. Live for the present. Look forward to the future.
Robot Fusion
"As our island of knowledge grows, so does the shore of our ignorance." John Wheeler
"[A] scientist looking at nonscientific problems is just as dumb as the next guy." Richard Feynman
"[P]etabytes of [] data is not the same thing as understanding emergent mechanisms and structures." Jim Crutchfield