Wednesday, December 23, 2009

Mapquest time algorithm

I have found Mapquest is usually pretty good about selecting the fastest route. However on my recent trip to the Outer Banks, I noticed an apparent problem with the suggested route just south of the Delaware Memorial Bridge. Mapquest suggested turning south off 295 onto 13 and then following 13 until it merges into 1. This is about 3.5 miles shorter than continuing on 295 to 95 and then turning south on 7 which also becomes 1. However the Mapquest route contains about 14 stoplights so it seems unlikely it is actually 3 minutes faster as claimed. I thought perhaps Mapquest was ignoring the stoplights although descriptions of their algorithm on the web claim they are accounted for (but not expected traffic delays). But when I tried Google directions it suggested the second route would be faster because Google estimated its time as 5 minutes less than Mapquest did while generating the same estimate for the shorter route. So perhaps Mapquest has incorrect data for the second route.

To confuse things more when the direction of travel is reversed Google also selects the first route after increasing the time for the second route by 3 minutes for no apparent reason. So perhaps there is something flaky about the data being used.

I used Deepwater, New Jersey and Middletown, Delaware as endpoints when experimenting.

No comments:

Post a Comment