A Hybrid ARIMA-ANN approach for optimum estimation and forecasting of gasoline consumption

Accurate estimation and forecasting of gasoline is vital for policy and decision‐making process in energy sector. This paper presents a hybrid data‐driven model based on Artificial Neural Network (ANN) and autoregressive integrated moving average (ARIMA) approach for optimum estimation and forecasting of gasoline consumption. The proposed hybrid ARIMA‐ANN approach considers six lagged variables and one forecasted values provided by ARIMA process. The ANN trains and tests data with Multi Layer Perceptron (MLP) approach which has the lowest Mean Absolute Percentage Error (MAPE). To show the applicability and superiority of the proposed hybrid approach, daily available data were collected for 7 years (2005–2011) in Iran. Although eliminating subside from gasoline price has led to appearing noisy data in gasoline consumption in Iran the acquired results show high accuracy of about 9427% by using the proposed hybrid ARIMA‐ANN method. The results of the proposed model are compared respect to regression’s models and ARIMA process. The outcome of this paper justifies the capability of the proposed hybrid ARIMA‐ANN approach in accurate forecasting gasoline consumption.