...
首页> 外文期刊>International Journal of Statistics and Applications >Estimation of Outliers in Periodic Processes for Different Underlying Mechanisms
【24h】

Estimation of Outliers in Periodic Processes for Different Underlying Mechanisms

机译:不同底层机制的周期性过程中的异常值估计

获取原文
           

摘要

This paper examines the performance of Additive Outlier (AO), Innovative Outlier (IO), Level Shift Outlier (LS) and Transitory Change Outlier (TC) based on autoregressive (AR), moving average (MA), and autoregressive moving average (ARMA) processes in the estimation of the magnitude of outliers for periodic processes. We consider the estimation of outlier magnitude under different generating mechanisms with special focus on its structure, bias and efficiency of the generating mechanism. We have empirically supported our analytical derivations that for ARMA, AR and MA processes, AO model would produce the best unbiased estimates compared with others and all processes and if , TC and LS models give a closer estimates precision as AO model. If the weight , IO may produce the same outlier as AO, TC and LS models. The analytical derivations in this study have shown that the magnitude of outliers depend respective weights of the models.
机译:本文研究了基于自回归(AR),移动平均(MA)和自回归移动平均(ARMA)的附加离群值(AO),创新离群值(IO),电平移位离群值(LS)和暂时性变化离群值(TC)的性能)估计周期性过程离群值的大小。我们考虑在不同的生成机制下的离群值的估计,特别关注其生成机制的结构,偏差和效率。我们有经验地支持我们的分析推导,即对于ARMA,AR和MA流程,与其他流程和所有流程相比,AO模型将产生最佳的无偏估计,如果TC和LS模型与AO模型相比,则估计精度更高。如果权重,则IO可能产生与AO,TC和LS模型相同的异常值。本研究的分析推导表明,离群值的大小取决于模型的权重。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号