Estimating gasoline demand in Iran using different soft computing techniques

Present study develops two scenarios to analyse gasoline consumption and makes future projections based on the particle swarm optimisation (PSO) and genetic algorithm (GA). The gasoline consumption is estimated based on the basic indicators of the population, gross domestic product (GDP), import, export, gasoline production and number of cars figures. Two different exponential and linear estimation models are developed for each scenario using PSO and GA methods. Developed models are validated with actual data, while future estimation of gasoline demand is projected between 2006 and 2030. For the best result (PSO - PGIE
_exponential
), the relative error average was 1.03%.