摘要
根据有关文献中把L1-回归问题归结为线性规划问题的有关结论,给出了最优解的条件及其等价形式,并讨论了L1-回归影响分析的两个问题:某一数据点的局部变动,对参数估计值的影响;某一数据点被删除后,对参数估计值的影响.
One of the problems studied on recursive model is the recursive influence analysis on the influence intensity of sample to statistical conclusion. Based on the results in [1], [2] which transform the L - regression problem into a particular kind of Linear Programming (LP), this paper gives conditions,for the optimal solution of the LP. Furthermore, by using the optimal conditions, two basic points of the influence analysis of the L -regression are considered, the influence of local change of one data point to parameter value: the influence of that after one data point is deleted.
出处
《湖北工业大学学报》
2007年第1期28-30,共3页
Journal of Hubei University of Technology
关键词
L1-估计
线性规划
影响分析
数据删除模型
L -estimator, linear programming, influence analysis, data deletion model