基于稳健回归的华北高校毕业生数预测  

The Prediction for North China College Graduates Number Based on Robust Regression

作　　者：谷闻聪 石晓冉[1] 王乔[1] 

机构地区：[1]北京师范大学数学科学学院,北京100875

出　　处：《数学的实践与认识》2013年第23期57-66,共10页Mathematics in Practice and Theory

基　　金：北京师范大学2011年度"北京市大学生科学研究与创业行动计划"

摘　　要：为促进和保障高等院校毕业生的就业,保证人才培养的数目和质量,提前准确地预测毕业人数尤为重要.使用经典最小二乘回归和稳健回归分析方法,选取可能的解释变量对华北五省的毕业人数进行分析.经过全子集法等方法筛选自变量.分别使用经典最小二乘方法和稳健方法建立模型,给出了两种方法的拟合结果对比和预测效果对比,结果表明稳健方法显著地提高了传统模型的拟合精度和预测精度.并预测了2018-2025华北五省的毕业生数,为有关部门制定相关政策提供可靠的数字依据.In order to promote college graduate employment status and guarantee the quality of graduate accurately, forecasting the graduate number is especially important. This article uses classical least-squares regression and the robust regression method to analysis the number of graduation in five provinces of northern China. The all-subsets regression method is used to screen initially explanation variables. The results of fitting and prediction effects for two methods are shown, which shows that the robust method has significantly improves the fitting precision and prediction precision of classical least-squares model. Finally forecasts graduate number of 2018-2025 of the five provinces in the next few years, which can provide policy- makers with valuable information.

关 键 词：高校毕业生数 多元回归分析 稳健回归分析 异常值 

分 类 号：G647.38[文化科学—高等教育学]