The idea I saw first in Marginal Revolution. This blogpost seems to be where the whole UHaul weighted graph idea came from. Dan Armstrong and Páll Hilmarsson
I took a list of the 34 most populous US cities (all over 500,000) from wikipedia. This is 1122 links in total. The 294 cities is 86142 total links. You only seem to be able to get containers not trucks from Honolulu, Hawaii.
This is a Complete graph where each edge has a length and a weight/capacity (price). some cities are really cheap to leave because enough people are moving (sinking) there that UHaul want to get the trucks back to the cities people are leaving (source)
The extreme costing trips are
Source | Destination | Price |
San Jose, CA | Washington, DC | 4237 |
San Francisco, CA | Baltimore, MD | 4188 |
San Jose, CA | Baltimore, MD | 4181 |
San Jose, CA | Washington, DC | 4132 |
Baltimore, MD | Washington, DC | 74 |
Washington, DC | Baltimore, MD | 79 |
The trips with the biggest difference between one way and another are by price
Source | Destination | Round dif | Round ratio |
San Jose, CA | Washington, DC | 2404 | 2.3 |
San Francisco, CA | Washington, DC | 2345 | 2.3 |
Philadelphia, PA | Portland, OR | 2213 | 3 |
and by ratio
Source | Destination | Round dif | Round ratio |
San Jose, CA | Las Vegas, NV | 580 | 4.1 |
San Francisco, CA | Las Vegas, NV | 608 | 4.1 |
Philadelphia, PA | Jacksonville, FL | 1301 | 3.9 |
The spreadsheet with these calculation is here. the code to work all this out is pretty raw but it is here.
I will come back to this later and work out Eigenvalue Centrality and maybe how distance relates to prices. Also it would be interesting to see if some places are summer sinks and some winter sinks in a few months time.
original scraper saved by the wayback machine
ReplyDeletehttps://web.archive.org/web/20060619101224/http://caesious.beasts.org/~chris/tmp/20050907/uhaulscrape