Implementasi Algoritma Long Short Term Memory (LSTM) dalam Memprediksi Konsumsi Energi Baru Terbarukan

Abstract

Increased global energy consumption is driving countries around the world to look for alternative energy sources. With the limited availability of fossil fuels, renewable energy is the right solution to address global energy needs. However, not all countries have developed the technology to implement the use of renewable energy as an energy source. Therefore, in this study, an attempt was made to build a renewable energy consumption prediction model with deep learning. The data used is energy consumption data in developing countries in Asia including Bangladesh, Brunei Darussalam, China, Philippines, India, Indonesia, Iran, Cambodia, Laos, Malaysia, Myanmar, Pakistan, Thailand, and Vietnam. Vietnam. The annual data includes Gross Domestics Percapita (GDP), Foreign Direct Investment (FDI), carbon emission levels, total fossil fuel consumption and total energy used. This data is processed through the preprocessing, training process with the LSTM model and testing, resulting a renewable energy consumption prediction model with the architecture and hyperparameters obtained from the training process results. The prediction model evaluation results are MAE with a value of 32,6503 and R2 of 0,9556.