Trip Fare Estimation Study from Taxi Routing Behaviors and Localizing Traces

It is hard for people to estimate the trip fare withthe dynamic traffic conditions and routing changes in a bigcity. In this paper, a comprehensive trip fare estimation study isconducted, which is based on taxi routing behaviors and localizingtraces. First, we build a model to investigate taxi drivers´ routing behaviors. Based on this model, we are able to analyze driverincome patterns. Second, we build a model to analyze trafficpatterns. Third, based on the above two models, we employ tuckermodel which is based on tenser decomposition to estimate taxifare of any route. To achieve better accuracy, we also introducesmoothing method into the model. Last, we test our method ona large-scale real-life data set.