Learning Dynamic Bayesian Networks for Analyzing Causal Relationship Between Macro-Economic Index

Dynamic Bayesian networks is a powerful tool in modeling multivariate stochastic processes. At present, the methods of learning dynamic Bayesian network structure have low efficiency and reliability and can not make certain the causal direction of all edges. In this paper, an high effective and reliable and practical method of learning dynamic Bayesian network structure is presented to find dynamic causal knowledge from data. Firstly, a maximal likelihood tree is built from data. Then a causal tree is obtained by orienting the edges of the maximal likelihood tree and the variables can be sorted in the light of the causal tree. Finally, a dynamic Bayesian network structure can be established based on the causal order of variables and local search & scoring method by finding father nodes of a node.