Cool. I did a traffic visualization from my apartment on Haight Street on New Years Eve and the first week of January this year. I just had a webcam with some software to detect movement in the middle of the intersection.
(My R skills aren't great, sorry it's not pretty. The camera also got blocked for awhile around day 5 so it's not exactly scientific)
That first cycle starts mid day on New Years Eve. You can see a big drop in traffic leading up to midnight, then a big increase soon after. Not surprising, but it's always a little gratifying to confirm a hypothesis.
Neat! Do you have the data posted anywhere? I'm not sure my code produces much of an improvement on yours, but you can get some nice results with ggplot2:
library(ggplot2)
x <- Sys.time() + 60 * (1:2400)
y <- rpois(2400, 50 * (sin(seq(-4 * pi, 4 * pi, len=2400)) + 1.1))
qplot(as.POSIXct(x), y, geom=c("point", "smooth"), xlab="time", ylab="count")
That's neat. I used to live in Ann Arbor, which is a quiet student town that gets very crowded on football weekends. It would be interesting to make up a game that makes users match unlabeled traffic graphs to the towns or neighborhoods where the graphs were generated.
http://tlrobinson.net/projects/haightcam/traffic.png
(My R skills aren't great, sorry it's not pretty. The camera also got blocked for awhile around day 5 so it's not exactly scientific)
That first cycle starts mid day on New Years Eve. You can see a big drop in traffic leading up to midnight, then a big increase soon after. Not surprising, but it's always a little gratifying to confirm a hypothesis.