Applications of neural networks in socio-economic modeling

In this paper we introduce a new artificial neural network (ANN) based approach for socio-economic modeling. This approach is most appropriate for modeling of real world problems when long time series data are available. Neural networks provide an alternative to other modeling paradigms (such as optimization, simulation, and econometrics) especially when the mathematical relationships between the dependent and independent variables are non-linear or not well defined. The paper presents an example of a socio-economic model with several environmental variables. This example is used as a basis for neural network model development. We present two different neural network approaches - the first ANN implementation is using a single large ANN while the second implementation is using a cascaded group of ANN sub-models. A comparison between the results given by the two neural network approaches is also presented in conjunction with statistical regression analysis of the variables.