Comparative Study of Elman Neural Network (ENN) and Neural Network Autoregressive With Exogenous Input (NARX) For Flood Forecasting

Abstract

Flood happen when a river ruptures its bank due to massive amount of water and the water spills out onto the land. Therefore, flood prediction system is necessary in order to alert the people about the incoming flood. Flood cannot be preventing but can take the precaution steps by using a precise technique to forecast the flood earlier. In this modern day, there are many models of flood prediction was introduced to the world. The accuracy of water level is very crucial for flood forecasting. By compare all the models, the most accurate technique can be discovered. In Malaysia, NARX and ENN are the flood prediction models that were frequently be used by researches. Thus, this paper proposed comparison between Elman Neural Network (ENN) and Neural Network Autoregressive with Exogenous Input (NARX) to specify which model are more accurate. Therefore, both of the model is using the same water level data which recorded in Kedah to ensure which model are more precise. Based on result obtain, Elman are more accurate than NARX since it have higher best fit and lower root mean square error (RMSE).