Curve fitting predication with artificial neural networks: A comparative analysis

Abstract

Artificial neural networks (ANN) is considered one of the most efficient methods in processing Big Data, they have a great potential in economics and engineering applications. The aims of this paper is to discuss the best method for forecasting time series by comparing the results of ANN and Box and Jenkins methods (BJ) or ARMA models. As well as finding the best curve fitting and forecasting for linear or semi linear model. In this paper uses 3 error indicators to measure the efficiency of forecasting for the forecasting performance. The most important conclusion of this paper Proved that artificial neural networks are more effective than Box-Jenkins method or ARMA models in solving time series. The results also proved that artificial neural networks are significantly improving errors in the results and this is the ambition of all researchers.