Stochastic Characteristics of Freeway Traffic Speed and Application to Short-Term Speed Prediction

Using a stochastic approach, this study examined and modeled the basic stochastic characteristics of freeway traffic behavior under a wide range of traffic conditions during peak periods and then applied the models to short-term traffic speed prediction. The speed transition probabilities were estimated from the real world 30-second speed data collected over a 6-year period on the 38-mile corridor of I-4 in Orlando, Florida. The cumulative negative/positive transition probabilities and expected values were derived from the transition probabilities and fitted using logistic and exponential models, respectively. The expected value associated with the most likely transition of speed was derived from the fitted models and used as the predicted speed. Each predicted speed was also associated with a probability value indicating the chance of observing such transition. The model performance was evaluated using Root Mean Square Errors (RMSE). Relatively small prediction errors of nearly 5 mph were observed. Also, the prediction performance was slightly affected by location, travel direction, and peak period. The study concluded that the behavior of freeway traffic can be modeled as a stochastic process, which can subsequently be applied to short-term traffic condition prediction during peak periods.