Time series smoothing by penalized least squares

被引:16
|
作者
Guerrero, Victor M. [1 ]
机构
[1] Inst Tecnol Autonomo Mexico, Dept Estadist, Mexico City 01000, DF, Mexico
关键词
mean square error; smoothness index; unobserved components;
D O I
10.1016/j.spl.2007.03.006
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
The time series smoothing problem is approached in a slightly more general form than usual. The proposed statistical solution involves an implicit adjustment to the observations at both extremes of the time series. The resulting estimated trend becomes more statistically grounded and an estimate of its sampling variability is provided. An index of smoothness is derived and proposed as a tool for choosing the smoothing constant. (c) 2007 Elsevier B.V. All rights reserved.
引用
收藏
页码:1225 / 1234
页数:10
相关论文
共 50 条