Abstract: The authors proposed a comprehensive simulation framework for real estate development by abstracting and modeling the self-learning behavior and interactions of the developers. Guided by the theory of ABM (Agent Based Model) and down-top routine, this framework was supported by data obtaining capacity of Remote Sensing and spatial analysis of GIS. The location choice model of developers employed multivariable regression model to achieve maximumutility, therefore developers bid for one land patch based on some competition mechanism. At last, experiments were taken in Beijing, and the authors analyzed the macro-spatial feature formed by micro behaviors of many local developers .

Key words: GIS, real estate development simulation, logistic regression model, location choice, ABM, urban simulation

摘要： 以遥感和地理信息系统作为数据获取和空间分析的技术手段，采用多主体建模方法，对房地产开发商的开发行为和相互作用过程进行抽象和模型化，提出一个自下而上的房地产开发模拟框架。采用多变量逻辑回归模型分别建立各类开发商的区位选择模型，各开发商通过竞标获取某个地块的开发权，并实施开发行为。以北京作为实验区，模拟研究了大量开发商通过微观相互作用形成某种宏观的房地产分布格局的复杂过程。

关键词: 地理信息系统, 房地产开发模拟, 逻辑回归模型, 区位选择, 基于主体的建模, 城市模拟