Optimization and Empirical Study of the STAR Models Parameter Estimation with Armijo-type Line Search

摘要：

　　The parameter estimation of the model optimization is a key part of system modeling.This paper presents a nonlinear estimation method for STAR model with Armijo-type inexact one-dimensional line search.Because of it only needs computing one order partial derivative and no stem demands for select initial point of objective function,it has the fast convergence speed and get more accurate results of parameters estimation in nonlinear STAR model and optimization this model.Empirical study show that Armijo inexact one-dimensional line search method estimating parameters with DFP and BFGS variable scale method optimized the STAR model and increased models fitting degree.

关键词： STAR model Parameter estimation Variable scale

作者: TIAN Kun WANG Qin CHANG Chunyan

作者单位: College of Mathematics & Information,China West Normal University,P.R.China,637002 School of Mathematics,Southwest Jiaotong University,P.R.China,610031

会议类型: 国际会议

会议名称: The 5th International Institute of Statistics & Management Engineering Symposium (第五届(2012)国际统计与管理工程学术研讨会(IISMES2012))

会议地点: 青岛

会议语种:英文

页码: 86-91

在线出版日期: 2012-07-20（万方平台首次上网日期，不代表论文的发表时间）