Taxi Driver Assistant – A Proposal for a Recommendation System (Internal Technical Report)

Authors

Abstract

Taxi is a flexible way of transportation; however, with the fast growth of urban areas, the task of finding new passenger quickly becomes a challenging task for taxi drivers. In this work, we propose a recommendation system to assist taxi drivers searching for the next pick-up. With an inference engine based on a naïve Bayesian classifier, we analyze taxi-GPS traces collected inLisbon, Portugal. An exploratory and predictive analysis is performed to better understand the apparent randomness of taxi movements, and the impact of features’ variation, namely, temporal periods, driver strategies or cell sizes, on the classifier.

Keywords

Urban mobility, taxi-GPS traces, spatiotemporal analysis, predictive analysis, naïve Bayesian classifier, recommendationsystem

Related Project

CityMotion (MITPortugal Program)