Identifying causal relationships in an urban information modeling framework

The modern city is a complex system with large amount of information. Discovering the hidden causal relationships among urban operational factors is a key research issue in city planning. Traditional approaches that concentrate on one aspect (e.g. transport) may have limitation when the city planning needs holistic analysis and an interoperable view of the city system. We introduce a systematic factor analysis approach for urban information by using Bayesian Network (BN) framework. We build a BN structure to present the causal relationships among city information factors based on integrated data sets. In addition, our approach is a dynamic modeling method because the learned BN structure will change in different time periods depends on the different service that the city provides. With structural leaning and real time monitoring, we perform Bayesian inference on the BN structure and the transformation of BN structure. We provide high level views to planners in the sense that identifying the potential cause or outcome of a city operational factor and examining the implementation of certain policies. Opinions from city planning experts may help us improve our model.