...
首页> 外文期刊>Quality Control and Applied Statistics >Improving estimation efficiency in quantile regression with longitudinal data
【24h】

Improving estimation efficiency in quantile regression with longitudinal data

机译:利用纵向数据提高分位数回归中的估计效率

获取原文
获取原文并翻译 | 示例
           

摘要

Quantile regression, in contrast to the least square regression provides a flexible way to determine the underlying relationships between response variables and the covariates. Many studies have been undertaken to improve the estimation efficiency for quantile regression in the recent years. Some studies used weights but are not varying as per the quantile levels. However if the weights can vary depending up on the heterogeneity in the data then it will be a more meaningful approach. Weights in the proposed approach are obtained from smoothened quantile estimator where quadratic inference function (QIF) (Qu et al., Ref. 1) is used to model the inverse of the correlation matrix of the conditional quantile score functions. The weights used here are from intra-subject conditional scores that may vary for different quantile levels. These weights can be used for composite estimation also to improve the estimation efficiency. The optimal quantile adaptive weights though are difficult to obtain when the data are longitudinal, this paper provides a useful approach to select weights in such condition without any computational problems. (23 refs.)
机译:与最小二乘回归相比,分位数回归提供了一种灵活的方法来确定响应变量和协变量之间的潜在关系。近年来,进行了许多研究来提高分位数回归的估计效率。一些研究使用权重,但未根据分位数水平变化。但是,如果权重可以根据数据的异质性而变化,那么它将是一种更有意义的方法。所提出方法中的权重是从平滑分位数估计器获得的,其中使用了二次推断函数(QIF)(Qu等人,参考文献1)来对条件分位数函数的相关矩阵进行逆建模。此处使用的权重来自对象内部条件得分,该得分可能因分位数不同而有所不同。这些权重可以用于复合估计,也可以提高估计效率。虽然当数据为纵向时难以获得最佳分位数自适应权重,但本文提供了一种在没有任何计算问题的情况下选择权重的有用方法。 (23篇)

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