数字地价模型的构建方法及参数选择

Construction Method and Parameters Setting of Digital Land Price Model

基于地统计内插构建数字地价模型是目前国内研究地价空间分布的常用手段。既往的研究大多侧重空间变异分析或不同模型结果的比较,较少探讨一个完整的建模过程及过程中一些重要参数如何科学选择与设置。文章以2008—2014年石家庄市主城区土地交易数据为例,借助ArcGIS、GS+软件平台,深入探讨了基于地统计学原理,利用Kriging插值的方法,构建数字地价模型的完整过程,并重点探究此建模过程中获取最优参数的方法。结果表明:基于地统计方法构建的数字地价模型,能够合理地模拟区域地价空间分布形态;并通过探索性空间数据分析、变异函数分析、不同模型拟合优度对比、交叉验证精度检验等手段,有效地保证了建模参数的科学性和准确性。

In view of the fact that the common means used to study the spatial distribution of land price in present China is based on geo-statistics to build digital land price model. The past studies mostly focus on the spatial variation analysis or the re- sult comparison between different models, but less on the discussion about the complete modeling process and how to set the pa- rameters scientifically. This paper uses the land transaction data of Shijiazhuang from 2008 to 2014 as samples and draws support from ArcGIS and GS＋ software platform to explore in depth the principle of geo-statistics, the method of using Kriging interpola- tion and the complete process of building a digital land price model, and also emphatically investigates how to obtain the optimal parameters in modeling. The study results show that the proposed method of building digital land price model based on geo-statis- tics more reasonably simulates the regional spatial distribution of land price; in the process of modeling, it can effectively guaran- tee the scientificity and accuracy of modeling parameters by means of exploratory spatial data analysis, variation function analysis, goodness-of-fit contrast of different model, and cross validation accuracy test, and so on.