Prediction Of Electricity Use Using A Website-Based Support Vector Machine Algorithm

Abstract

This study aims to create an electrical load prediction system using the Support Vector Machine algorithm to be able to predict future electrical loads. This study also finds out what parameters can reduce the error rate of predictions using Particle Swarm Optimization. Then everything is packaged into a website using the flask framework. The results of testing the parameters of the Support Vector Machine algorithm on the electricity usage prediction system, the lowest error values obtained are MAE, MSE, RMSE on the parameters of the PSO optimization results, the SVR parameter value is C = 1; Gamma=8.3; Epsilon=0.001; produces an error value, MAE=0.00829921; MSE=0.00602241; RMSE= 0.0776042. Keywords—Support Vector Machine, Particle Swarm Optimization, Prediction, Penggunaan Energi Listrik