Forecasting interval time series using a fully complex-valued RBF neural network with DPSO and PSO algorithms

Abstract

Interval time series prediction is one of the most challenging research topics in the field of time series modeling and prediction. In view of the remarkable function approximation capability of fully complex-valued radial basis function neural networks (FCRBFNNs), we set out to investigate the possibility of forecasting interval time series by denoting the lower and upper bounds of the interval as real and imaginary parts of a complex number, respectively. This results in a complex-valued interval. We then model the resulted complex-valued interval time series via a FCRBFNN. Furthermore, we propose to evolve the FCRBFNN by using particle swarm optimization (PSO) and discrete PSO for joint optimization of the structure and parameters. Finally, the proposed interval time series prediction approach is tested with simulated interval time series data as well as real interval stock price time series data from the New York Stock Exchange. Our experimental results indicate that it is a promising alternative for interval time series forecasting.