基于支持向量机的人口失衡风险预警方法研究

发布时间：2018-03-17 00:19

本文选题：支持向量机　切入点：人口失衡　出处：《广西民族大学》2012年硕士论文　论文类型：学位论文

【摘要】：2011年,我国“十二五”规划纲要提出“促进人口长期均衡发展”的战略目标,使得人口失衡风险预警成为当今人口领域研究的新课题。人口失衡风险的表现形式复杂,理论上可分为240种不同类型,具有内生性、扩散性等特征。人口失衡风险问题的复杂性,增大了人口失衡风险预警的难度,用传统的统计评估方法或系统评估方法均难以取得满意效果。本文研究基于支持向量机的人口失衡风险预警方法。支持向量机技术适于处理小样本、非线性和高维数据,并且具有优良的泛化性能,切合人口失衡风险预警的技术需求。首先,在剖析人口失衡风险点的基础上,抽取出能表征人口失衡风险的指标,构建人口失衡风险预警指标体系并制定指标预警阈值。其次,将支持向量机、粗糙集、决策树等智能计算方法引入人口风险预警领域,构造模块化的基于支持向量机的人口失衡风险预警模型。该模型的一个模块是一组标准支持向量机和粗糙集支持向量机,用于实现总体风险警度预报；另一个模块是一组决策树支持向量机,用于实现风险的分项评估,根据分项评估结果可准确定位样本风险点。第三,针对难分类样本,构造由分项风险评估结果预测总体风险警度的粗糙集核验模型。以分项风险评估结果作为条件属性,以总体风险警度作为决策属性,构建决策表,利用粗糙集算法生成决策规则,对难分类数据进行分类,以此结果核验原粗糙集支持向量机分类结果的准确性。这一算法也可以在其他涉及分项知识的专业领域当中应用。
[Abstract]:In 2011, the outline of the 12th Five-Year Plan put forward the strategic goal of "promoting the long-term balanced development of population", which made the risk early warning of population imbalance become a new topic in the field of population. The manifestation of population imbalance risk is complex, which can be divided into 240 different types in theory, which has the characteristics of endogenicity and diffusivity. The complexity of population imbalance risk makes it more difficult to predict population imbalance risk. It is difficult to obtain satisfactory results by using traditional statistical evaluation methods or systematic evaluation methods. In this paper, the population imbalance risk early warning method based on support vector machine (SVM) is studied. Support vector machine (SVM) is suitable for processing small sample, nonlinear and high dimensional data, and has excellent generalization performance. First of all, on the basis of analyzing the risk point of population imbalance, we extract the index which can represent the risk of population imbalance, construct the early warning index system of population imbalance risk and establish the warning threshold. The intelligent computing methods such as support vector machine, rough set and decision tree are introduced into the field of population risk early warning. A modular population imbalance risk early warning model based on support vector machine (SVM) is constructed, one of the modules of the model is a set of standard support vector machines and rough set support vector machines, which can be used to predict the overall risk alarm. Another module is a set of decision tree support vector machines, which can be used to realize the sub-assessment of risk. According to the results of sub-evaluation, the risk points of samples can be accurately located. A rough set verification model is constructed to predict the overall risk alarm from the results of the itemized risk assessment. The decision table is constructed with the result of the itemized risk assessment as the conditional attribute and the total risk alarm as the decision attribute. Rough set algorithm is used to generate decision rules to classify difficult classification data, which verifies the accuracy of classification results based on rough set support vector machine. This algorithm can also be used in other specialized fields involving sub-knowledge.
【学位授予单位】：广西民族大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：C924.2;TP18

【参考文献】