A novel approach for urbanization level evaluation based on information entropy principle: A case of Beijing

• Novel urbanization level evaluation model is proposed based on information entropy.
• Beijing’s urbanization level has ever kept increasing during 2005–2012.
• Contributions of economic and urban–rural development to Beijing changed most.
• Poor coordination of urban–rural development is largely existed in Beijing.
• We further discuss the objectivity and flexibility in choosing indicator network.

Urbanization level evaluation (ULE) is an important scientific basis for guiding urban managers to make decisions. By introducing information entropy to describe the interactions between all indicators, a holistic structural parameter ξξ, its dynamic equation and self-organizing feature map simulation technique are derived to describe the structural evolution of the indicator network. In this way, a novel ULE model is universally proposed. Then, we use the model to assess the evolutionary urbanization level of Beijing during 2005–2012. We calculate structural parameter ξξ values of the indicator network with 35 microscopic indicators as nodes. The results show Beijing’s urbanization level has ever kept increasing. Large increase of ξξ values in 2008 and 2012 represented significant improvements of urbanization level in these two years, while a rapid adjustment of urbanization development occurred in 2010. Five meso-scopic subsystems as urban construction, economic development, social development, ecological environment and urban–rural development affected Beijing’s urbanization level in different ways. The radar chart of the model shows the contributions of economic development and urban–rural development to Beijing’s urbanization changed most, while poor coordination of urban–rural development largely existed. By showing Beijing’s ULE based on two analytical ways, we further discuss the objectivity and flexibility in choosing indicator network. Finally, beyond the application case, we discuss the universality and superiority of the new model.