摘要
构建自然资源评价指标集,不仅要剔除相关性高的指标,还要保留解释性强的重要指标。但当前对指标解释性的评估多是基于主观分析进行,以评价人自身的标准去衡量指标,缺乏客观判断。该文融合指标解释性和独立性,提出最优化模型,并引入复相关系数和关系指数,通过松弛迭代计算,对候选集指标进行优化调整,直至代价函数取得极大值或趋于稳定,从而得到最优指标集合。通过3个实验证明:该方法在规避多重共线性和评价结果稳定性方面优于相关系数-变异系数法。
To construct a natural resource evaluation indicator set,it is not only necessary to eliminate highly correlated indicators,but also to retain important explanatory indicators.However,the current evaluation of indicator interpretability is mostly based on subjective analysis,using evaluators'own standards to measure indicators,lacking objective judgment.In this paper,the interpretation and independence of the indicators are integrated,an optimization model is proposed,and the complex correlation coefficient and relation index are introduced.Through relaxation iterative calculation,the candidate set indicators are optimized and adjusted until the cost function reaches a maximum value or becomes sTab.,and the optimal set of indicators is obtained.Finally,three experiments show that the proposed method is superior to the correlation coefficient-coefficient of variation method in avoiding multicollinearity and evaluating the stability of results.
作者
杨晶
张鹏林
YANG Jing;ZHANG Penglin(School of Remote Sensing and Information Engineering,Wuhan University,Wuhan 430079,China)
出处
《测绘科学》
CSCD
北大核心
2023年第9期240-251,共12页
Science of Surveying and Mapping
基金
广东省自然资源厅2022年度科技项目(GDZRZYKJ2022001)
国家重点研发计划项目(2022YFC3006305)。
关键词
综合评价
评价指标筛选
指标体系
自然资源
comprehensive evaluation
evaluation indicator screening
indicator system
natural resource
