Calculating Route Probability from Uncertain Origins to a Destination

Abstract

Uncertainty in location information can affect the results of network-based route calculations to a high degree. In this study, a routing scenario is analysed where the destination is known, but the location of the point of origin can only approximately be described as “somewhere inside a polygon”. Using the concrete example of car-driving football fans arriving at a game, an approach is proposed to compute and describe the probability of them taking specific routes from their home county to the stadium. A set of candidate points of origin is created and shortest paths to the destination calculated. The observed frequency of an edge being included in a route allows inferring a routing probability for each edge. Several methods to derive a set of candidate points of origin are presented and discussed, ranging from purely geometrical to geographically weighted approaches. Our results show that the differences between the methods in determining the points of origin produce only slightly different probabilities, i.e., neither advantages nor drawbacks are to be expected from using a purely geometrical approach.