Rainfall Forecasting in Malaysia Using Hybrid LSTM Models

Yeo, Xi Yong (2022) Rainfall Forecasting in Malaysia Using Hybrid LSTM Models. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Yeo Xi Yong_FullText.pdf Restricted to Registered users only Download (1MB)

Abstract

Forecasting rainfall is a task that must be carried out for every country in this world. It is important because the amount of rainfall not only will affect the life of humans but also other living organisms. A large amount of rainfall for a continuous period may cause flooding while a small amount or even no rainfall may cause drought. Hence, if the amount of rainfall can be forecasted accurately, then humans can plan and take action early to overcome these problems. However, due to the non linear and complicated nature of rainfall, the task of forecasting rainfall is not that easy. In this paper, hybrid models of LSTM which are ARIMA-LSTM and EMD-LSTM are constructed to forecast the monthly rainfall at Gua Musang, Kelantan, Malaysia. The results obtained from the hybrid models are compared to the single ARIMA and LSTM model. The results are evaluated by performance metrics such as Root Mean Squared Error (RMSE) and Coefficient of Determination (R Squared).The results show that both of the hybrid models perform better than the single models which means lower error and higher correlation. Between the two hybrid models, EMD-LSTM outperforms the ARIMA-LSTM model by giving the lowest RMSE and highest