Using generalized linear models and time series models to forecast gasohol consumption in Thailand

Abstract

This paper proposed two methods in statistical modeling to forecast gasohol consumption as alternative energy in Thailand.  The generalized linear models (GLMs) and time series models were applied. The first method was used to investigate the factors that impact response variables, consumption of gasohol 91, gasohol 95, gasohol E20, and gasohol E85. Also, the first method was to take among of factors to forecast all of response variables. The second method was used to forecast gasohol consumption in the further. The important results were ethanol price impacting to all of response variables: higher ethanol price making smaller consumption of gasohol 91 and gasohol 95. On the other hand, higher ethanol price making higher consumption of gasohol E20 and gasohol E85. However, it has been found in this study that forecast precision in terms of MAPE of GLMs models were less than time series models.