China has become the largest energy consumer worldwide, and it is important to study the energy intensity to realize the sustainable development goal of China. This paper focuses on investigating the influential factors of China's energy intensity using provincial-level panel data from 1985 to 2012. More specifically, we try to identify which factor is relatively more important to pay attention to. A novel approach based on evolutionary computation is proposed to intelligently mine the intrinsic relations between observed phenomena and to let the important factors automatically emerge from the discovered nonlinear models. However, due to China's vast territory and significant heterogeneities, this approach may fail to examine some detailed or hidden information when analyzing the country as a whole. Instead, we concentrate on the provincial level because the provinces play vital roles in reducing energy intensity in China. From our analytical results, the main findings are as follows: (1) the Total Population is the most important influential factor across China's provinces, while the Energy Price Index has the least impact; and (2) the provinces could be naturally classified into four categories based on the primary factors emerged from data, and such classification could reveal more about the true underlying features of each area. (C) 2015 Elsevier Ltd. All rights reserved.