Assessing the probability of arriving on time using historical travel time data in a road network

Abstract

This paper introduces a novel prediction model for assessing travel time reliability in a network considering alternate paths. Here the travel time reliability is defined as the probability of arriving within a given time budget. For that purpose historical travel time data provided by the BMW Group is used. When planning a trip, there are usually multiple alternate routes leading to the destination. At the time the trip is scheduled to start there is real-time information on the expected travel time available. Hence a path can then be chosen, based on the suggestion of a routing algorithm. When scheduling a trip for the future, there is no real-time information available yet and the decision which path to take is not ultimate. Thus it is crucial to not only base the travel time prediction on one path, but rather look at the data for each alternate path. In this paper the choice for which path to take is represented in a decision tree. The probabilities for the branches of the decision tree are derived from the historical travel time distributions. The travel times are represented as random variables with corresponding probability density functions. These density functions are convolved using Fourier Transformation in order to obtain the density function of the total path. By including the total network instead of only relying on the travel time information of one particular path it is shown that the travel time reliability increases.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 5 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.