arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0711.1918stat.ME
收藏

修正的残差信息准则

The Residual Information Criterion, Corrected

Chenlei Leng

摘要:

Shi和Tsai(JRSSB,2002)提出了一个有趣的残差信息准则(RIC)用于回归中的模型选择。他们的RIC的动机是最小化真实模型和候选模型的残差似然之间的Kullback-Leibler差异的原则。然而,我们表明,在这个原则下,RIC总是会选择完整的(饱和)模型。因此,在定义信息标准时,剩余似然不适合作为差异度量。我们解释了为什么会这样,并提供了纠正的剩余信息标准作为补救措施。

英文摘要:

Shi and Tsai (JRSSB, 2002) proposed an interesting residual information criterion (RIC) for model selection in regression. Their RIC was motivated by the principle of minimizing the Kullback-Leibler discrepancy between the residual likelihoods of the true and candidate model. We show, however, under this principle, RIC would always choose the full (saturated) model. The residual likelihood therefore, is not appropriate as a discrepancy measure in defining information criterion. We explain why it is so and provide a corrected residual information criterion as a remedy.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑