Comparative Analysis of Backpropagation and Radial Basis Function Neural Network on Monthly Rainfall Prediction

Abstract

Rainfall plays an essential role in the overall growth of any country. It is very crucial to forecasting the rainfall accurately and on time due to the highly erratic nature of weather conditions. Its accurate and predetermined forecast can prevent many natural hazards and is helpful to the tourists for traveling beautiful destinations of the world. Artificial Neural Network is functionally analogous to the human brain. Earlier, due to its innate capability of performing highly complex calculations, researchers began to develop an interest in designing a computer model which can work in a way similar to the human brain. Under this study, we have applied the two commonly known techniques of artificial neural network backpropagation and radial basis function neural network for predicting rainfall. Furthermore, mean square error (training and testing) and accuracy are the performance indices used for the comparative analysis of the two models. The motive of this thesis is to outline the working of these two approaches for accurate prediction of rainfall and advantages of using these methods over the other ANN techniques. In this study, we have also analyzed the prediction results of the two algorithms and determined which of the two is better regarding performance results. Further, the ideas are suggested to improve the prediction by combining the ANN methods with several other algorithms. A GUI software has been created in MATLAB for easy prediction of future rainfall data.