基于空间自相关和时空扫描统计量的聚集比较分析  

COMPARATIVE ANALYSIS OF AGGREGATION DETECTION BASED ON SPATIAL AUTOCORRELATION AND SPATIAL-TEMPORAL SCAN STATISTICS

作　　者：王培安[1] 罗卫华[2] 白永平[1] 

机构地区：[1]西北师范大学地理与环境科学学院,兰州730070 [2]中国航天科技集团公司第四研究院四〇一所,西安710025

出　　处：《人文地理》2012年第2期119-127,共9页Human Geography

基　　金：国家自然科学基金项目(40771054);高等学校博士学科点专项科研基金联合资助课题(20106203110002);西北师范大学知识与科技创新团队项目(nwnu-kjcxgc-03-50)

摘　　要：聚集是区域经济研究中的重点问题之一,而聚集定位问题又是深入分析聚集需要解决的首要问题。由于聚集具有高度的尺度敏感性,采用空间自相关方法分析时,尺度选择容易受研究者主观判断的影响,而且空间自相关方法也未考虑聚集的时间特征。与之相比,Kulldorff等学者提出的扫描统计量方法表现出了明显的优势。研究探索性地选用浙江省各市、县工业从业人口的聚集问题,从尺度选择、尺度转换和时空融合三个方面,比较了空间自相关和时空扫描统计量方法在探测聚集问题上的差异性,进而证实了时空扫描统计量方法不仅有效解决了人为选择尺度的偏倚问题,实现了尺度推绎、转换的自动化,而且更加有利地融合了立体、动态、多尺度的时空分析优势。The problem of aggregation economy has been always one of the hot and focuses of the regional economic research,and the positioning of aggregation is a prerequisite for continuing in-depth study in aggregation analysis.Using different methods,the scope of the aggregation phenomenon is different.Because aggregation is high sensitivity in scale,the influence of scale must be fully considered in using the spatial analysis methods.Using spatial autocorrelation methods to detect aggregation,the scale of choice is often susceptible to subjective judgments of the researchers;there is the issue of possible with selection bias,so the spatial weight problem has been controversial.For spatial autocorrelation analysis,the spatial weight matrix creation is the most difficult problems.Using different spatial weight matrix will get different results.In addition,the aggregation is obviously space-dependent as well as time-dependent since different aggregation takes place within different time and space,which is neglected by the spatial autocorrelation method.This is a concrete manifestation of the difference in time and space.So space-time analysis is necessarily a method which must be selected.At this point,the scan statistics method which was developed by Kulldorff and other scholars demonstrated a unique advantage.The exploratory research will use data about industrial population in some cities and counties of Zhejiang from 2000 to 2009.In particular,by comparing the space-time scan statistics method and a simple spatial autocorrelation method,which effectively introduces a time variable,this approach allows researchers to solve the problem a long time,but this approach can also increase the accuracy of the results.This shows that it is not only effective measure to solve the problem of artificial selection,to achieve the Scale extrapolation and automatic conversion,and more favorable mix of three-dimensional,dynamic,multi-scale analysis.

关 键 词：尺度矩阵 空间自相关 时空扫描统计量 聚集 

分 类 号：F119.9[经济管理—国际贸易]