摘要
最小一乘估计是人们最常用的回归方法之一,因为其回归结果受奇异点的影响较小,从而受到人们越来越多的关注,鉴于此方法所构造模型的非光滑性,进而增加了其计算的难度.针对不同观测结果及需求,将最小一乘模型转化成不同的线性规划模型,利用相应的求解软件进行求解.并针对不同情况对结果进行了灵敏度分析,从而找出了影响结果的因素.
Least squares estimation is a kind of commonly regression method. In this method, the regression results are less affected by the singualaxity, so it is concerned more and more by many scholars. Given the smoothness of the constructed model in the way. Thus there are some difficulties in he calculation. According the different observations and demand. Articles least absolute deviation model transforms to different linear model and Solutes it, then conductes a sensitivity analysis, thus finds the factors of fluencing the results.
出处
《数学的实践与认识》
CSCD
北大核心
2011年第11期227-231,共5页
Mathematics in Practice and Theory
关键词
最小一乘估计
线性规划
灵敏度分析
目标函数
Least Absolute Deviation Estimator(LAD)
linear programming
Sensitivity analysis
objective function