Construction of Demand Forecast Model of Tokyo Taxi Based on Probe Data Analysis

Abstract

We construct a decision support model that can help taxi drivers dispatch their vehicles appropriately based on machine learning by utilizing probe data of taxis in Tokyo. Traditionally, taxi dispatch has relied on the driver's experience and intuition. The number of customers depends on the knowledge gained through many years of experience. For examples, many taxis are sometimes waiting for customers, and sometimes many customers wait in a long queue for a taxi. In addition, there are differences in the transport distance depending on the location. However, in a given situation, not all drivers know places with high demand. Therefore, it is desirable to build a model, which is easy to understand for drivers, that enables efficient acquisition of customers regardless of their experience. In this situation, we propose an analysis model that supports driver's decision based on taxi probe data.