lidegithub / neural-network-project Goto Github PK

This project forked from zzclynn/neural-network-project

Time series forecasting have been applied in many areas like environment, industry, economy and finance. People use several methods like Linear regression (LR), exponential smoothing (ES), autoregressive integrated moving average (ARIMA) to predict linear time series. Previous studies have shown that artificial neural network can model non-linear pattern of time series data. “The neural network structures and training procedures will have a great impact on forecasting performance” (Tang and Fishwick). However in real world, time series data consist of linear and non-linear patterns. Purwanto et al suggest combining Linear Method and Non Linear Method for more accurate time series forecasting. This project deals with examining the performance of neural network combining with linear method.