基于支持向量机的人的全面发展评价模型及省份实证

根据“坚持以人为本,树立全面、协调、可持续的发展现,促进经济社会和人的全面发展”的科学发展观的内涵,通过人的全面发展评价指标的海选、筛选和理性分析构建了人的全面发展综合评价指标体系.建立了基于支持向量机(Support Vector Machines,SVM)的人的全面发展评价模型,并对我国2006年典型的14个省级行政区人的全面发展状况进行实证分析.本文的创新与特色一是通过高斯核支持向量机把评价指标空间映射到高维特征空间,解决了人的全面发展评价影响因素非线性赋权问题.克服了现有评价方法均采用线性加权方式计算评价结果、不能表达指标与评价结果间的真实关系的缺陷.二是通过正交设计确定支持向量机训练样本的输入数据,并通过AHP确定训练样本的权重,得到指标训练样本的评价结果作为训练样本的输出.解决了在没有训练样本输入和输出情况下如何确定指标训练样本的问题.这就解决了在缺乏训练样本输入、输出数据情况下如何应用支持向量机进行回归,以得到评价结果的问题.三是通过人口累积比重、收入累积比重、通货膨胀等可获得数据指标计算准基尼系数和准国民幸福指数,解决了基尼系数和国民幸福指数的间接测算问题,进而解决了现阶段因统计数据缺失、而无法进行省级行政区人的全面发展评价的问题.

The Human All-Round Development Evaluation Model Based on Support Vector Machines and an Empirical Research on 14 Provinces in China

The Scientific Outlook on Development study includes four basic principles in economic and social development: balanced,people-oriented,comprehensive,and coordinated.The core principle of Scientific Outlook on Development is people-oriented.The important aspect of implementing Scientific Outlook on Development is to establish the human all-round development evaluation system that reflects the core principles of Scientific Outlook on Development.Establishing the human all-round development evaluation system based on Scientific Outlook on Development principles has a great significance on the scientific development of China. According to the connotation of Scientific Outlook on Development,this paper constructs a comprehensive human all-round development evaluation index system through selection,screening,and rational analysis of human all-round development indicators.This paper establishes human all-round development evaluation model based on support vector machines,and analyzes the development on fourteen provinces in China as of 2006. This paper is organized as follows.Section 1 contains the introduction.Section 2 constructs an index system of human all-round development evaluation based on the Scientific Outlook on Development.Section 3 describes the construction of evaluation model of human all-round development evaluation based on support vector machines.Building on the index system and the evaluation model,section 4 discusses empirical research on human all-round development of fourteen provinces in China as of 2006.This section discusses survey data used in this study,the evaluation process,and the results of this study.Finally,section 5 contains the conclusion. This paper has three aspects of contributions.Firstly,this paper maps the evaluation indicator space into the higher-dimensional feature space through support vector machines using the Gauss kernel.This solves the nonlinear weighting problem of human all-round development evaluation and changes the disadvantage of the existing evaluation method that adopts linear weighting method without considering the nonlinear relationship between the indicators and the evaluation results. Secondly,this paper designs input indicators of training set for support vector machines through an orthogonal model.At the same time,this paper designs training set for support vector machines through the product of inputs and weights.The training set of support vector machines is determined afterwards.The weight of training set is designed through AHP.This solves the problem of determining the training set without input and output.Also,this changes the comprehensive evaluation model of human all-round development based on support vector regression without training data. Thirdly,this paper calculates Gini Coefficient and Gross National Happiness by using the cumulative proportion of population,the cumulative proportion of income,inflation,and other indicators.This design solves the indirect calculation problem of Gini Coefficient and Gross National Happiness,and changes the existing research that is unable to evaluate the regional human all-round development because of statistical data flaws.