The factors affecting on consumption of some petroleum product in Thailand using general linear model and generalized linear model

Abstract

The objective of this research was to investigate forecasting consumption of some petroleum products in Thailand. Two methods of statistical modeling, general linear models (GLM) and generalized linear model (GLMs) were applied. The GLM compared with GLMs in which response variables are the consumption of gasohol 95, gasohol 91, gasohol E20, and gasohol E85. The explanatory variables were gasoline 95 price, gasohol 91 price, gasohol 95 price, LPG price, NGV price, ethanol price, exchange rates, and world crude oil prices. The secondary data from many resources in the time period Jan 2010 to Dec 2015 were collected for analysis. The factor analysis was applied to reduce some-covariate variables that are high relationship. The study found that the GLMs with gamma distribution was more appropriate than GLM when response variable is not normality. In case of the response variable is normally distributed both of them give well the same result.