Abstract: Considering spatial factors, the present study took geographical space effects into STIRPAT model to reveal the stochastic impacts by regression on P (population), A (affluence) and T (technology). Total energy consumption was used as indicators of environmental pressure, and population density, GDP and the proportion of secondary industry were adopted to represent population, affluence and technology, respectively. Geographical weighted regression (GWR) model was applied to estimate the elasticity of driving factors in 16 districts of Beijing. It was shown that the change of driving factors was not balanced in different districts. But, the changes in every district exhibited certain rule. Therefore, it is possible to formulate characterized regulation and control policies of energy production for different districts.

Key words: energy consumption, driving factors, STIRPAT model, geographically weighted regression