Learning a Causal Model from Household Survey Data by Using a Bayesian Belief Network

Abstract

A Bayesian belief network (BBN) is a modeling and knowledge-representation structure used in artificial intelligence that consists of a graphical model depicting probabilistic relationships among variables of interest. This graphical model is a valuable tool for representing the causal relationships in a given set of variables. Because the number of possible BBNs for a given data set is exponential with respect to the number of variables, learning a BBN from data is a difficult and resource-consuming task. A greedy algorithm that automatically constructs a BBN from a data set of cases obtained from a household survey was implemented. The resulting BBN shows the dependencies among key variables that are associated with the trip-generation process.

Get full access to this article

View all access options for this article.

References

Russel

, and Norvig

Artificial Intelligence. A Modern Approach. Prentice Hall, Englewood Cliffs, N.J., 1995.

Heckerman

A Tutorial on Learning with Bayesian Networks. Technical Report MSR-TR-95-06. Microsoft Research, Redmond, Wash., 1996.

Cooper

G. F.

, and Herskovitz

A Bayesian Method for the Induction of Probabilistic Networks from Data. Machine Learning, Vol. 9, 1992, pp. 309–347.

Robinson

R. W.

Counting Unlabeled Acyclic Graphs. In Lecture Notes in Mathematics 622: Combinatorial Mathematics ( Little

C. H. C.

, ed.), Springer-Verlag, New York, 1977.

Ramoni

, and Sebastiani

On the Use of Bayesian Networks to Analyze Survey Data. Research in Official Statistics, Vol. 4, No. 1, 2001, pp. 53–64.

Transportation Research Circular E-C008: Transport Surveys: Raising the Standard–Proceedings of an International Conference on Transport Survey Quality and Innovation. TRB, National Research Council, Washington, D.C., 2000. TRB.org/trb/publications/ec008/ec008.pdf.