Taxi demand analytics and driver's recommendation system: Algorithm and evaluation

Finding more passengers is the most important problem for newbie taxi drivers. Even experienced drivers who are not familiar with particular areas. They have to find more passengers as soon as possible because each cruise takes more energy cost and creates opportunity loss. This research introduced the integration of a recommendation system for taxi drivers, which can guarantee the best profits them. To clarify how the system works, after dropping off passengers, three recommendations are displayed to the drivers, i.e., hubs, POIs and routes as digital maps on their mobile devices. Each recommendation relates to the demand for taxis in that area and depends on the time-of-day analyzed by data mining algorithms. We implemented the developed system and showed the performance of the real-world taxi system supported by All Thai Taxi. The drivers' performances were improved by up to 10.5% after the use of the recommendation system.