Analysis of Long Short-Term Memory and Support Vector Regression Methods in Forecasting Electric Energy Sales: Case Study

Septiawan, Adi Harjo
Fauzi, Umar

Journal La Multiapp
Vol. 6 No. 2 (2025): Journal La Multiapp

This study aims to predict the sales of electrical energy of PT PLN (Persero) Greater Jakarta Distribution Unit by using machine learning methods, specifically Long Short-Term Memory (LSTM) and Support Vector Regression (SVR). The data used includes electrical energy sales trends from 2016 to 2023 as well as external data from the Central Statistics Agency (BPS), which includes economic and demographic factors that affect energy demand, such as economic growth, population, and seasonal factors. LSTM was chosen for its ability to handle long-term dependencies in time series data, while SVR was used as a comparison to other regression methods. The resulting model is expected to provide more accurate predictions and be useful for PT PLN in planning the distribution of electrical energy efficiently. This research also contributes to the development of the application of machine learning in forecasting, which is growing in various sectors, including the energy sector, to improve operational efficiency and data-based decision making.