Quantitative models for forecasting vehicle fuel prices in Thailand

In this research, quantitative models for forecasting vehicle fuel selling prices at a gas station in Thailand are investigated. Four types of gasoline include Gasoline 95, Gasoline 91, Gasohol 95, and Diesel. The data are drawn from the year 2002 to 2008. Time series and regression methods are used in this research. A composite method is also proposed using three different approaches in assigning the weights. Firstly, the weights minimize the variance of the combined error. Secondly, the weights are taken from a regression. Thirdly, the weights are all equal. The results such as the best model for prices of Gasoline 91 and Gasohol 95 is the combined method of using regression in obtaining the weights are presented. © 2008 IEEE.