基于Lasso稳健马田系统的相对贫困识别方法  

Identification of relative poverty based on Lasso-based robust Mahalanobis-Taguchi system

作　　者：陈闻鹤 程龙生[1] 常志朋 周涵婷[1] CHENWenhe;CHENG Longsheng;CHANG Zhipeng;ZHOU Hanting(School of Economics&Management,Nanjing University of Science and Technology,Nanjing 210094,China;School of Business,Anhui University of Technology,Maanshan 243002,China;Key Laboratory of Multidisciplinary Management and Control of Complex Systems of Anhui Higher Education Institutes,Anhui University of Technology,Maanshan 243002,China)

机构地区：[1]南京理工大学经济管理学院,南京210094 [2]安徽工业大学商学院,马鞍山243002 [3]安徽工业大学复杂系统多学科管理与控制安徽普通高校重点实验室,马鞍山243002

出　　处：《系统工程理论与实践》2022年第2期527-544,共18页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(71673001);江苏省研究生科研与实践创新计划项目(KYCX21_0356);安徽省高校优秀青年人才支持计划重点项目(gxyqZD2017040);安徽省普通高校重点实验室开放基金重点项目(CS2020-ZD02);安徽省高校人文社会科学基金重大项目(SK2021ZD0034)。

摘　　要：2020年后我国扶贫工作重点将转向解决相对贫困问题,然而解决相对贫困的前提和基础是识别相对贫困.由于相对贫困数据具有"高噪性、不平衡性、相对性、多维性"等特点,现有的单一收入维度的贫困识别方法难以适用,因此提出构建一种基于Lasso稳健马田系统(Lasso-based robust Mahalanobis-Taguchi system,Lasso-RMTS)的相对贫困识别方法.该方法通过Lasso、稳健马氏距离和马田系统三方法融合,使得不仅可以对"不平衡性"和"比较性"的贫困数据进行识别,还可以对"高噪性"和"多维性"的相对贫困数据进行降维、降噪.同时,将稳健马氏距离转换为脱贫指数(poverty alleviation index,PAI),可以更加直观地反映相对贫困程度.实例数据验证表明,Lasso-RMTS识别相对贫困的精准度高于马田系统和其他传统分类方法.After 2020,the focus of China’s poverty alleviation work will turn to solving the problem of relative poverty.However,the premise and basis of solving relative poverty is to identify relative poverty.Due to the characteristics of relative poverty,such as "high noise,imbalance,relativity and multi-dimension",it is difficult to apply the method of relative poverty identification with single income dimension.Therefore,the Lasso-based robust Mahalanobis-Taguchi system(Lasso-RMTS) is constructed to identify relative poverty.Through the integration of Lasso,robust Mahalanobis distance(RMD),and Mahalanobis-Taguchi system(MTS),this method can not only identify the poverty data of imbalance and relative comparability,but also reduce the dimension and noise of the relative poverty data.At the same time,the RMD can increase the sensitivity of relative poverty data by transforming it into the poverty alleviation index(PAI).The data shows that Lasso-RMTS can accurately and efficiently identify relative poverty,and its performance is better than MTS and other traditional methods of classification.

关 键 词：马田系统 脱贫指数 相对贫困识别 Lasso 稳健马氏距离 生计可持续框架 

分 类 号：F303.2[经济管理—产业经济]