A Short-Term Prediction Model for Forecasting Traffic Information Using Bayesian Network

Currently the traffic information services of Telematics have had high qualities due to easy collection of the real-time traffic information through intelligent transport system (ITS). In this work, a short-term prediction model is proposed for forecasting the traffic information. The Bayesian network is used for each link with some casual nodes which can affect road situations in the future. In addition, a joint probability density function of the Bayesian network is obtained by assuming Gaussian Mixture Model (GMM) which utilizes training data set. To validate the precision of our model we conducted various experiments with two measures, one is an index as root mean square error (RMSE) and the other is travel time which takes three kinds of shortest paths for given paths. Our model provides less than 8 value of RMSE and the travel time of dynamic shortest path has more than 85% correlation with the real traffic data.