Model Peramalan Konsumsi Bahan Bakar Jenis Premium di Indonesia dengan Regresi Linier Berganda

Abstract

Energy consumption forecasting, especially premium, is an integral part of energy management. Premium is a type of energy that receives government subsidy. Unfortunately, premium forecastings being performed have considerable high error resulting difficulties on reaching planned subsidy target and exploding the amount. In this study forecasting was conducted using multilinear regression (MLR) method with ten candidate predictor variables. The result shows that only four variables which are inflation, selling price disparity between pertamanx and premium, economic growth rate, and the number of car, dictate premium consumption. Analsys on the MLR model indicates that the model has a considerable low error with the mean absolute percentage error (MAPE) of 5.18%. The model has been used to predict 2013 primium consumption with 1.05% of error. The model predicted that 2013 premium consumption was 29.56 million kiloliter, while the reality was 29.26 million kiloliter.