Quantile regression in varying coefficient models
Quantile regression in varying coefficient models
复制标题
DOI：
10.1016/s0378-3758(03)00110-1
复制
发表时间：
2004-03-01
影响因子：
0.9
通讯作者：
Honda, T
中科院分区：
文献类型：
--
作者：
Honda, T
本文通过估计系数来处理不同系数模型中条件分位数的估计。变系数模型是为了减轻维数灾难而提出的流行模型之一。先前关于变系数模型的工作直接或间接地处理条件均值。然而，分位数本身可以在没有矩条件的情况下定义，并且绘制几个条件分位数将使我们比仅绘制条件平均值更好地理解数据。特别是，我们通过局部 L 回归估计变化系数来估计条件中位数。 (C) 2003 Elsevier B.V. 保留所有权利。
This paper deals with the estimation of conditional quantiles in varying coefficient models by estimating the coefficients. Varying coefficient models are among popular models that have been proposed to alleviate the curse of dimensionality. Previous works on varying coefficient models deal with conditional means directly or indirectly. However, quantiles themselves can be defined without moment conditions and plotting several conditional quantiles would give us more understanding of the data than plotting just the conditional mean. Particularly, we estimate the conditional median by estimating varying coefficients by local L, regression. (C) 2003 Elsevier B.V. All rights reserved.