摘要
针对目前GDP空间化存在的精度偏低、运算复杂等问题,提出了一种基于机器学习算法的产业结构空间化方法。文章以成渝地区双城经济圈为研究区域,将各区县产业产值空间化为1 km×1 km的精度,利用GBDT和RF算法,基于多源数据,对产业产值进行预测。分析结果发现:预测模型的决定系数可以达到0.91;GBDT模型性能表现优于RF模型;成渝地区双城经济圈内第一产业高值区集中在乌江、涪江、沱江一带,低值区位于南部平原;第二产业、第三产业集中在成都、重庆两大城市,存在“极化”现象。该研究成果对于推动成渝地区双城经济圈的产业协同发展、建设现代化产业体系、优化产业布局具有重要的指导意义,同时也为产业结构空间化研究提供了一种新的思路。
A new method based on machine learning algorithm for the spatialization of industrial structure is presented for the problems that the current method of GDP spatialization has low precision and is complicated.This article takes the Chengdu-Chongqing economic circle as the research area and spatializes the value of each industry into 1 km×1 km accuracy.Using GBDT and RF algorithms,the output value of industries is predicted based on multi-source data.The results show that:the coefficient of determination of the prediction model based on multi-source data can reach 0.91;the GBDT model performs better than the RF model in terms of performance;the high-value areas of the primary industry in the Chengdu-Chongqing economic circle are concentrated on both sides of the Wujiang,Fujiang,and Tuojiang rivers,while the low value areas are located in the southern plain;the secondary and tertiary industries are concentrated in the two major cities of Chengdu and Chongqing,and there is a phenomenon of polarization.The research results have important guiding significance for promoting the coordinated development of industries in the Chengdu-Chongqing economic circle,building a modern industrial system,and optimizing industrial layout,while also provide a new approach for industrial spatialization research.
作者
韩成龙
朱创业
HAN Chenglong;ZHU Chuangye(College of Geography and Planning,Chengdu University of Technology,Chengdu 610059,China)
出处
《遥感信息》
CSCD
北大核心
2024年第4期125-133,共9页
Remote Sensing Information
关键词
机器学习
GBDT
夜间灯光
产业结构空间化
成渝地区双城经济圈
machine learning
GBDT
night light
spatialization of industrial structure
Chengdu-Chongqing economic circle
