A Mokov Decision Process Approach to Optimizing Waiting for Taxis
2018
Taxi services play a critical contribution for the public transportation system in large cities. However, we usually find that our intrinsic experience is often mistaken to wait for taxis smoothly, especially for the passengers travelling to one unfamiliar city. This will greatly affect the user experience of taxi services. Therefore, how to recommend a good waiting place for one passenger is a meaningful problem, and the existing large-scale trajectory data just provides us with more opportunity. Current researches focus on one-time recommendation for passengers, such as recommending the waiting location closest to the current passenger, the one with minimum waiting time or hot spots, etc. However, most recommended strategies recommend only one location to the current passenger without considering the recommendation failure that waiting for no taxis. In response to this deficiency, we propose a Mokov Decision Process (MDP) approach for recommend a sequence places for the current waiting passenger with the constraints of time and distance by solving the value functions. This ensures the maximum probability for waiting for a taxi throughout the process and the overall optimization of the waiting process. A case study and several experimental evaluations on a real taxi dataset from a major city in China show that our proposed MDP approach is in line with the actual situation and reduce greatly the waiting time for passengers and improve the user experience of taxi services.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
14
References
2
Citations
NaN
KQI