Who's onlineThere are currently 0 users and 0 guests online.
|
Visualization ProjectOne of my current personal projects (out of very many) is coming up with a way to visualize some of the data that we have at work (the C-SPAN Archives). The first thing I've been working on is doing a social network type graph where the nodes are people that have appeared on C-SPAN, and the edges connect nodes where those people have appeared together. Hopefully with the end result you will be able to see groups of people and be able to tell that this group is all Senators, or this group are all teachers who consistently appear at conferences together. I started working on a Flash application from scratch, but recently found the Flare project with Flex libraries for Flash that already do 90% of anything I could have hoped to accomplish. After surprisingly little work (setting up a webservice, playing with defaults..) it is already a point where it is starting to look interesting. Here is the current graph for the up to 75 people who have appeared on C-SPAN with Kurt Vonnegut the most. Because I have imported the most recent (up to I think 01/2009) data from Metafilter's cortex's InfoDump into a web-hosted MySQL database I was able to set up a web service for that too and the only change from the C-SPAN visualization to the Metafilter one is the address of the xml. One thing I've done is removed the edge of the person for whom the data is being returned. It cluttered up the graph where everyone was attached to the same person. The oddity that arises here is that Vonnegut and Matthowie are now just floating around.. Here's what it looked like before I did this: There's still a lot to do. There should be all kinds on configuration options for users, the ability to add more nodes of interest, links to webpages, etc. It would also be good to have colors for the C-SPAN app to identify what kind of person is shown (Govenor, Spouse, Teacher..) For the Metafilter graph I am currently ignoring the contact/contacted by disticntion, but it wouldn't be too hard to make a directed graph with different colors to show this. We also don't have contact types in the InfoDump, but with this information you could see how many contacts per person are spouses (a surprisingly high number). Grr Early BirdsI'm going to have to agree with Senator Graham here: Nothing good happens before 9:30. Second attempt at a random album cover.Random Wikipedia article: Van_Bergen's Regiment of Militia Result: Random Stephen Jay Gould talk from 1996This would, of course, apply to all other sports and competitive environments and explain why nobody really watches professional sports any more. Beyond the clip he ties the presentation together into a primordial argument for nihilism. Hurf Durf Data Eater
http://www.ammurapi.com/files/hurfDurf.tar.bz2 Needed: What it does: Used for: Error handling? Naahh.. Good luck! The initial importing takes a couple hours.. the db is possibly over-indexed. The data to be downloaded is a hundred MB or so. The DB created is around 750 MB. Thanks for the meat. All cubed.I lurk and occasionally post on metafilter. One of the admins has a Markov thingy A creation based on my ramblings:
Album CoversBack in January I found this cool meme: It wasn't until today that I've tried it.
Sports StatsHere's the basketball stats page that I'm working on. The link goes to the IU/Western Michigan game. I haven't been working on it too much lately because we are in conference play and the MAC doesn't publish play-by-play sub in/out information. Without that we only have the points graph. IntroDuring the day I do software projects. The rest of the time I try to do anything but software.. |
Navigation |