摘要
目前,中国失业率统计存在一定局限,不利于准确及时地反映劳动市场的就业变动,大数据技术的快速发展为中国失业率统计提供新的发展视角.基于网络搜索数据,文章从5种常用的预测方法中筛选出最优的支持向量机回归模型,对中国季度失业率进行了预测研究.研究表明,基于网络搜索数据预测的失业率能够比官方数据更早地反映失业趋势的变化,预测失业率与修正后的失业率水平接近,能够为政府部门提供中国失业状况的政策预警.
At present, there are some limitations in China's unemployment rate statistics, which is not conducive to accurately and timely reflect the employment changes in labor market. The rapid development of large-scale data technology pro- vides a new development perspective for China's unemployment statistics. Based on the network search data, this paper selects the best support vector machine regression model from five commonly used forecasting methods, and forecasts the quarterly un- employment rate in China. The results show that the unemployment rate can reflect the change of the unemployment trend sooner than the official data. The predictedunemployment rate is close to the revised unemployment rate, which can provide the government with the policy warning of the unemployment situation in China.
出处
《系统科学与数学》
CSCD
北大核心
2017年第2期460-472,共13页
Journal of Systems Science and Mathematical Sciences
基金
国家社科基金重大项目(2015YZD08)
国家社科基金项目(14CRK019)
国家自然科学基金项目(71573034)
辽宁省教育厅项目(LN2016JD020)
中国博士后科学基金(2016M601318)资助课题
关键词
失业率预测
大数据
网络搜索数据
支持向量机.
Unemployment statistics, big data, network search data, support vectormachine.