Bayesian Network Methods for Traffic Flow Forecasting with Incomplete Data

Abstract

Traffic flow forecasting is an important issue in the field of Intelligent Transportation Systems. Due to practical limitations, traffic flows recorded can be partially missing or unavailable. In this case few methods can deal with forecasting successfully. In this paper two methods based on the concept of Bayesian networks are originally proposed to cope with this matter. A Bayesian network model and a two-step Bayesian network model are constructed respectively to describe the causal relationship among traffic flows, and then the joint probability distribution between the cause and effect nodes with its dimension reduced by Principal Component Analysis is approximated through a Gaussian Mixture Model. The parameters of the Gaussian Mixture Model are learned through the Competitive EM algorithm. Experiments show that the proposed Bayesian network methods are applicable and effective for traffic flow forecasting with incomplete data.