北京大学学报(自然科学版)

基于自组织特征映射网络的全国地级市城市地价区域分类研究

高阳,赵瑞娜,阿杉,李双成   

  1. 地表过程分析与模拟教育部重点实验室, 北京大学城市与环境学院, 北京 100871;
  • 收稿日期:2009-05-07 出版日期:2010-07-20 发布日期:2010-07-20

SOFM-Based Classification for Land Price of City in China

GAO Yang, ZHAO Ruina, A Shan, LI Shuangcheng   

  1. Laboratory for Earth Surface Processes, Ministry of Education, College of Urban and Environmental Sciences, Peking University, Beijing 100871;
  • Received:2009-05-07 Online:2010-07-20 Published:2010-07-20

摘要: 地级市是我国经济发展相对迅速的区域, 也是土地供应重组、交易活跃的重点区域。以2005年我国土地出让面积、土地平均价格、地区生产总值、地区生产总值增长率和 固定资产投资等 5 个变量作为聚类指标, 构建自组织特征映射( SOFM) 人工神经网络模型, 将我国282 个地级市分为高地价发达区、低地价发达区、高地价欠发达区和低地价欠发达区共4 个类型区域, 并对每个类型区的土地价格和社会经济发展状况做出分析讨论。SOFM 模型聚类结果与客观实际较为吻合, 效果良好。结果表明, 自组织特征映射 网络对于地级市土地地价的区域差异具有良好的表征能力。

关键词: 城市土地, 经济增长, 自组织特征映射网络(SOFM), 城市分类

Abstract: Cities at prefectural level(area cities) are not only high-speed economic developing areas, but also the key areas of land supply, reorganization and active transaction. Five variables such as area of land transfer, average land prices, GDP, growth rate of GDP, and fixed assets investment are used to develop a self-organizing feature map(SOFM) artificial neural network model. The results show that 282 area cities in China are divided into the four categories: developed area of high land prices, developed area of low land prices, underdeveloped area of high land prices, underdeveloped areas of low land prices. According to the results, the characteristics of each region are analyzed and the current development situation is discussed. Classification results match the objective reality very well, indicating SOFM-based classification method is an alternative approach in research of socio-economic development.

Key words: urban land, economic growth, self-organizing feature map(SOFM), urban classification

中图分类号: