Abstract
Recently, the market for location-based services (LBS) has been rapidly expanding. As an LBS value-added service, a concierge service provides users with a minimum total cost route using multiple modes. One of the core features in a concierge service is a travel time forecast which undoubtedly affects decisions by users as to the selection of modes, routes and time of travel. Multimodal travel time forecasts thus become an essential part of providing LBS. Many researchers have endeavored to develop reliable travel time forecasting models. Most of these studies, however, have shown general forecasting performances without carefully evaluating performances of models with actual time data. The purpose of this paper is to implement a transit travel time forecasting model and evaluate the performance. The nonparametric regression model, developed in an earlier study has been implemented and evaluated using real-time transit data.
