Modeling and Forecasting Residential Natural Gas Demand in IRAN

The main focus of this paper is to provide an appropriate mathematical model to predict the natural gas demand for the next six months in the household sector using technical method regardless of influential variables. For this purpose, the most important and most widely used modeling methods of natural gas demands were used as modeling options and four different candidate families were analyzed based on 234 months of historical data of actual consumption and model parameters were estimated using appropriate methods. Then, by using the precision indicators such as average absolute error, average absolute percentage error, waste diffraction model and inequality coefficient of Thiel time series, predicted for a period SARIMA (1,1,2) (1,1,0)12 accepted as the most appropriate fitness function were Identified and of the next 6 months. The results showed that the proposed method has minimum residual in terms of MAE, MAPE, TIC and error variance.