Multi-polynomial higher order neural network group models for financial data and rainfall data simulation and prediction

Abstract

Multi-Polynomial Higher Order Neural Network Group Models (MPHONNG) program developed by the author will be studied in this thesis. The thesis also investigates the use of MPHONNG for financial data and rainfall data simulation and prediction. The MPHONNG is combined with characteristics of Polynomial function, Trigonometric polynomial function and Sigmoid polynomial function. The models are constructed with three layers Multi-Polynomial Higher Order Neural Network and the weights of the models are derived directly from the coefficents of the Polynomial form, Trignometric polynomial form and Sigmoid polynomial form. To the best of the authors knowledge, it is the first attempt to use MPHONNG for financial data and rainfall data simulation and prediction. Results proved satisfactory, and confirmed that MPHONNG is capable of handling high frequency, high order nonlinear and discontinuous data.

Date of Award	2001
Original language	English