Harmony Search algorithm-based gasoline consumption modeling for Indonesia

Abstract

Indonesia is facing the discrepancy problem of the realized amount of gasoline consumption from the planned consumption. The discrepancy has burdened Indonesian national’s budget, because these unplanned imported amount of gasoline has increased government payment to the importer. In order to decrease this discrepancy, we intend to build a more accurate estimation model of the gasoline consumption in Indonesia. The objectives of this research are to develop a robust and accurate model to forecast future gasoline consumption and to provide an attractive alternative for gasoline consumption modeling by applying a metaheuristic approach. Utilizing general socioeconomic variables that can be retrieved from public data easily, we apply the Harmony Search (HS) algorithm to develop a model of gasoline consumption in Indonesia. The HS algorithm works to optimize the weight factors within the proposed mathematic models. Three Harmony Search algorithm-based models are proposed in this study: HS-Quadratic, HS-Linear, and HS-Exponential. Variables used in these models are Indonesian GDP, Indonesian population, and the total number of passenger cars and motorcycles in Indonesia. The HS-based Exponential model outperforms the other models. Its Mean Absolute Percentage Error (MAPE) reached 10.46%, while HSbased Linear model and HS-based Quadratic model obtained 12.15% and 18.514%, respectively.