Title
Predicting Driver Destination Using Machine Learning Techniques
Abstract
In this paper we present a method for predicting the driver's destination with 96% accuracy. Knowing the driver's destination has many useful applications in traffic safety, traffic mobility, and influencing driver behavior. Furthermore, a software application that can predict the driver's destination can reduce the burden on the driver from manually entering the destination address on small-screen mobile devices. Current methods for predicting driver destination do that by predicting the driver's route. Those methods result in 72% accuracy if relying only on GPS traces. By providing accurate map data, the accuracy of prediction in current methods can reach 98% in best case scenarios that were tested on one subject. We propose an alternate approach that separates the destination prediction problem from the route prediction problem. We provide ten subjects, of varying driving patterns, with a Smartphone equipped with a GPS tracking software application and ask them to turn on the tracking application for all their trips for a period ranging between 2 and 9 weeks. We apply an algorithm to detect Origins and Destination from GPS traces and feed that into several machine learning algorithms to model the destination. Given the current position of the driver, the position at which the driver was 5 min prior, the time of day, and the day of the week; the algorithm provides a prediction of the destination with a 1000m resolution. The Decision Tree with Pruning algorithm proves to be the most accurate resulting in an average 96 +/- 1.72 % accuracy of prediction across all subjects.
Year
DOI
Venue
2013
10.1109/ITSC.2013.6728224
2013 16TH INTERNATIONAL IEEE CONFERENCE ON INTELLIGENT TRANSPORTATION SYSTEMS - (ITSC)
Keywords
Field
DocType
tracking,algorithms,decision trees,learning artificial intelligence,global positioning system
Time of day,Decision tree,Ask price,Simulation,Mobile device,Software,Ranging,Global Positioning System,Artificial intelligence,Engineering,Assisted GPS,Machine learning
Conference
ISSN
Citations 
PageRank 
2153-0009
5
0.47
References 
Authors
5
2
Name
Order
Citations
PageRank
christian manasseh150.47
Raja Sengupta21468258.78