基于多指标和SOS-SVM的失业预警研究  

Research on Unemployment Early Warning Based on Multiple Indicators and SOS-SVM

作　　者：刘芳 LIU Fang(Xingzhi College, Xi'an University of Finance and Economics, Xi'an, Shanxi 710038, China)

机构地区：[1]西安财经大学行知学院,陕西西安710038

出　　处：《微型电脑应用》2020年第6期112-114,118,共4页Microcomputer Applications

摘　　要：为实现高精度的失业预警和为劳动保障部门进行失业预警提供理论参考和方法依据,从市场维度、职业维度、企业维度、行业维度、区域维度和群体维度等6个维度构建出一套失业预警指标体系,并针对支持向量机模型性能受惩罚参数和核参数的影响,运用共生生物搜索算法对SVM模型参数进行优化选择,建立基于多指标和SOS-SVM的失业预警模型。与PSO-SVM和SVM对比发现,SOS-SVM进行失业预警具有更高的准确率。In order to achieve high precision of unemployment warning,a set of unemployed early-warning index system is established.The system can be used to provide theoretical reference and method basis for the department of labor security.The system is built with six dimensions i.e.,market dimension,professional dimension,enterprise dimensions,industry dimension,social dimension and communal dimension.Because the penalty parameter and kernel parameter of support vector machine may be influenced,symbiosis biological search algorithm is used to optimize the parameters of the SVM mode.Based on multi-index and SOS the unemployed early-warning model is constructed by SVM.Compared with PSO-SVM and SVM,it is found that SOS-SVM has a higher accuracy for unemployment warning.

关 键 词：支持向量机 共生生物搜索算法 失业预警 粒子群算法 评价指标 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]