A Stochastic Volatility Modeling Approach to Account for Uncertainties in Travel Time Reliability Forecasting

摘要

Travel time effectively measures freeway traffic conditions. Easy access to this information providesthe potential to alleviate traffic congestion and to increase the reliability in road networks. However,it is still a challenging task to model and estimate travel time, as traffic often reveals irregularfluctuations. Traditional point prediction methods often underestimate the irregular fluctuations andprovide results that are prone to be uncertain. To capture travel time fluctuations and uncertaintiesassociated with prediction, this paper proposes the Autoregressive Integrated Moving Average-Stochastic Volatility (ARIMA-SV) model that generates the expected value of travel time (a pointvalue) as well as a prediction interval. An advanced Monte Carlo Markov Chain estimation method isemployed to fit the stochastic volatility model. The experiment results based on travel time datacollected from Bluetooth detectors along an I-95 segment in Connecticut suggest that the proposedARIMA-SV model out performs the ARIMA-GARCH model in both congested and non-congestedsituations. The proposed method has shown its advantages in capturing traffic fluctuations and has thepotential to disseminate more reliable traffic information to travelers through Advanced TravelerInformation Systems (ATIS).