Quantitative Modelling in Economics with Advanced Artificial Neural Networks

AbstractIn this paper, authors present a new approach in forecasting economic time series - application of artificial neural networks. Authors apply feed forward artificial neural network of the RBF type into the process of forecasting the financial data. Except for the standard RBF, authors also test their own new versions of this neural network combined with other techniques of the ML. These models represent new and more advanced version of the standard neural network. Authors add the evolutionary approach into the ANN and also combine the standard algorithm for adapting weights of the ANN with an unsupervised clustering algorithm called K-means. Finally, all of these methods are compared and contrasted with standard (statistical) approach on real economic data to show the potential of using artificial neural network in modelling economic variables