Get our free extension to see links to code for papers anywhere online!

Chrome logo  Add to Chrome

Firefox logo Add to Firefox

Predicting passenger origin-destination in online taxi-hailing systems

Oct 17, 2019
Pouria Golshanrad, Hamid Mahini, Behnam Bahrak

Share this with someone who'll enjoy it:

Because of transportation planning, traffic management and dispatch optimization importance, the passenger origin-destination prediction has become one of the most important requirements for intelligent transportation systems management. In this paper, we propose a model to predict the origin and destination of travels which will occur in the next specified time window. In order to extract meaningful travel flows we use K-means clustering in four-dimensional space with maximum cluster size limitation for origin and destination. Because of large number of clusters, we use non-negative matrix factorization to decrease the number of travel clusters. We also use a stacked recurrent neural network model to predict travels count in each cluster. Comparing our results with other existing models show that our proposed model has 5-7% lower mean absolute percentage error (MAPE) for 1-hour time window, and 14% lower MAPE for 30-minute time window.

* 23 pages, 20 figures 

   Access Paper Source

Share this with someone who'll enjoy it: