Jack May
2008-06-05 21:15:58 UTC
Interesting research in the news today about how much people travel in
normal life. The answer is most people don't travel much beyond home and
work.
http://news.wired.com/dynamic/stories/S/SCI_CELL_PHONE_TRACKING?SITE=WIRE&SECTION=HOME&TEMPLATE=DEFAULT&CTIME=2008-06-05-01-28-03
"Researchers secretly tracked the locations of 100,000 people outside the
United States through their cell phone use and concluded that most people
rarely stray more than a few miles from home."
"It also yielded somewhat surprising results that reveal how little people
move around in their daily lives. Nearly three-quarters of those studied
mainly stayed within a 20-mile-wide circle for half a year."
"The scientists would not say where the study was done, only describing the
location as an industrialized nation. The study was based on cell phone
records from a private company, whose name also was not disclosed."
"Study co-author Cesar Hidalgo, a physics researcher at Northeastern, said
he and his colleagues didn't know the individual phone numbers because they
were disguised into "ugly" 26-digit-and-letter codes. They started with 6
million phone numbers and chose the 100,000 at random to provide "an extra
layer" of anonymity for the research subjects, he said."
I used the data from the San Jose Mercury news and the best I could
interpret in the at times badly worded Wired article.
A curve fit of the data says Cumulative probability =
1.2559*Radius^(-0.7277) R^2 = .9964
Radius is the distance in miles that a person predominantly stays within.
Not surprisingly gas station locations, gas sales location and road surface
have similar power law distributions as people mobility with power
coefficients of -.77. -.79 and -.83
http://www.pnas.org/cgi/reprint/104/17/7301?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&fulltext=Geoffrey+West&searchid=1&FIRSTINDEX=0&resourcetype=HWCIT
There is also a study with a similar result where dollar bills were tracked
to see how people move around.
http://people.esam.northwestern.edu/~brockmann/index_assets/Brockmann_EPJ2008.pdf
Both distributions are of course a power law cumulative distribution as are
almost all complex system real world statistics. Real world statistics
are not even remotely Gaussian ( bell curve). Mean and variance are almost
meaningless on power law statistics which is why efficiency data for
different modes of transportation are essentially worthless. I don't even
have much confidence in the R^2 value calculated by Excel given above,
although the curve fit looks fairly good.
normal life. The answer is most people don't travel much beyond home and
work.
http://news.wired.com/dynamic/stories/S/SCI_CELL_PHONE_TRACKING?SITE=WIRE&SECTION=HOME&TEMPLATE=DEFAULT&CTIME=2008-06-05-01-28-03
"Researchers secretly tracked the locations of 100,000 people outside the
United States through their cell phone use and concluded that most people
rarely stray more than a few miles from home."
"It also yielded somewhat surprising results that reveal how little people
move around in their daily lives. Nearly three-quarters of those studied
mainly stayed within a 20-mile-wide circle for half a year."
"The scientists would not say where the study was done, only describing the
location as an industrialized nation. The study was based on cell phone
records from a private company, whose name also was not disclosed."
"Study co-author Cesar Hidalgo, a physics researcher at Northeastern, said
he and his colleagues didn't know the individual phone numbers because they
were disguised into "ugly" 26-digit-and-letter codes. They started with 6
million phone numbers and chose the 100,000 at random to provide "an extra
layer" of anonymity for the research subjects, he said."
I used the data from the San Jose Mercury news and the best I could
interpret in the at times badly worded Wired article.
A curve fit of the data says Cumulative probability =
1.2559*Radius^(-0.7277) R^2 = .9964
Radius is the distance in miles that a person predominantly stays within.
Not surprisingly gas station locations, gas sales location and road surface
have similar power law distributions as people mobility with power
coefficients of -.77. -.79 and -.83
http://www.pnas.org/cgi/reprint/104/17/7301?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&fulltext=Geoffrey+West&searchid=1&FIRSTINDEX=0&resourcetype=HWCIT
There is also a study with a similar result where dollar bills were tracked
to see how people move around.
http://people.esam.northwestern.edu/~brockmann/index_assets/Brockmann_EPJ2008.pdf
Both distributions are of course a power law cumulative distribution as are
almost all complex system real world statistics. Real world statistics
are not even remotely Gaussian ( bell curve). Mean and variance are almost
meaningless on power law statistics which is why efficiency data for
different modes of transportation are essentially worthless. I don't even
have much confidence in the R^2 value calculated by Excel given above,
although the curve fit looks fairly good.