interns

Notes regarding project status and info

10 posts • Page 1 of 1

Postby uhtu » Tue Aug 22, 2006 7:07 pm

we'd like to congratulate the wigle^tm lab interns who have spent the last six months updating our data filtering/ranking/triangulating algorithms as their "summer" project.
this should result in some minor signal increases in the very noisy data set that is wigle. seti@home eat your heart out.

Postby pejacoby » Tue Aug 22, 2006 10:27 pm

Uh guys, the interns appear to have eaten a few hundred thousand data points on the way.....<<burp>>...we're back to less than 7 million ?

6,977,155 to be exact at 5:27 PM CDT.

Postby bobzilla » Thu Aug 24, 2006 2:33 am

Aye, the number goes down from time to time as better methods of fingerprinting garbage are developed. This time it pushed under a milestone.

Postby whitedice » Thu Aug 24, 2006 3:38 am

Normally the stats page seems pretty accurate, but I was wondering if something is not correct with these #s?

New unique networks today: 293
New networks today with location: 225
Files parsed today: 434
Queued files waiting to process: 346

And a "Last Post" sorted showing only 2 people have posted new files on 8/23 which will become 8/24 in approx. 20 minutes (EST)?

According to the stats, hratch and mungewell have combined to post 434 files and found 293 new networks? right? Also, I could have sworn Mark571 had something like 58+k new APs found this month a couple of days ago (I'm pretty sure I was third again, something like 3 months in a row). But I guess I could have been dreaming about that last part... scary. I just have this weird feeling something isn't right with the #s is all.

Postby Tech^CF » Thu Aug 24, 2006 8:14 am

My jaw dropped and I'm quite shocked to see the stats whitedice mentions
Files parsed today: 434
Queued files waiting to process: 384
Guess I'll have to wait some time to see if I'll make it into the first stat page today

Postby bobzilla » Fri Aug 25, 2006 1:14 am

Due to batching, the numbers seen there might not be what you would expect during a large processing run.

Postby bobfunland » Fri Aug 25, 2006 12:28 pm

Aye, the number goes down from time to time as better methods of fingerprinting garbage are developed. This time it pushed under a milestone.
Damn, I guess driving around with Fake AP running in the backseat is frowned upon here huh?

Postby stefanb » Fri Aug 25, 2006 1:54 pm

I noticed waaay too many APs stacked on top of each other here
http://www.wigle.net/gps/gps/Map/online ... 8533782958

It is a city centre, APs are dense in reality, but not that dense. More likely it was that someone entered location of the city instead actual GPS readout or used some position rounding.

I guess the algorithm could be further improved to detect those

Postby uhtu » Fri Aug 25, 2006 3:09 pm

what you're seeing in that webmap snippet is a rendering artifact, and not actually eleventy points stacked on one another. a query (or use of a real client) shows only one of each offending network.
and yes, the algorithms are pretty much in continuous improvment, even when it makes things worse :-)

Postby i_do_dew » Sat Sep 02, 2006 11:29 pm

I noticed waaay too many APs stacked on top of each other here
http://www.wigle.net/gps/gps/Map/online ... 8533782958

It is a city centre, APs are dense in reality, but not that dense. More likely it was that someone entered location of the city instead actual GPS readout or used some position rounding.

I guess the algorithm could be further improved to detect those

Actually since my home scanning station never moves, it alsways has the same location. about 20 peers drive by on the highway every day!

10 posts • Page 1 of 1

Return to “WiGLE News”

Who is online

Users browsing this forum: Bing [Bot] and 3 guests