RE: "USING NUMERICAL METHODS TO DESIGN SIMULATIONS: REVISITING THE BALANCING INTERCEPT".

RE: "USING NUMERICAL METHODS TO DESIGN SIMULATIONS: REVISITING THE BALANCING INTERCEPT".
复制标题

回复:“使用数值方法设计模拟:重新审视平衡截距”。

DOI:
10.1093/aje/kwac083
复制
发表时间:
2022
影响因子:
5
通讯作者:
Ross,RachaelK
Ross,RachaelK
中科院分区:
医学2区
文献类型:
--
作者:
Zivich,PaulN;Ross,RachaelK

文献摘要

参考文献

被引文献

相似文献

在《华尔街日报》上,Robertson等人(1)重新审视了Rudolph等人最近提出的逻辑模型的“平衡拦截”(2)。正如Robertson等人所说。请注意,模拟数据最好是模拟真实数据的某些特征。平衡截取背后的动机是允许调查人员在数据生成模型中指定特定于协变量的参数,同时使用截取来将生成的协变量的边际值操纵到期望的水平。对于像Logistic回归这样的非线性模型,不存在一般的闭式解,但已经提出了解析近似。正如Robertson等人所说。举例而言,解析近似可能并不总是产生预期的结果。相反,在Logistic回归的情况下,数值近似被证明产生了预期的结果。正如Robertson等人所说。在他们的结论中强调,Logistic模型平衡拦截只是可以使用数值近似的一个例子。在这封信中,我们通过提供一个解决几个不同分布的平衡截取的说明性示例以及相应的R和Python代码来强调这一点。
In the Journal, Robertson et al.(1) revisit the “balancing intercept” for logistic models recently proposed by Rudolph et al.(2). As Robertson et al. note, it is desirable for simulated data to mimic certain features of real data. The motivation behind the balancing intercept is to allow an investigator to specify covariate-specific parameters in a data-generating model, while using the intercept to manipulate the marginal value of the generated covariate to a desired level. For nonlinear models, like logistic regression, there does not exist a general closed-form solution, but analytical approximations have been proposed (2). As Robertson et al. illustrate, analytical approximations may not always produce the desired results. In contrast, numerical approximations were demonstrated to produce the desired results in the case of logistic regression. As Robertson et al. emphasize in their conclusion, the logistic-model balancing intercept is only one example where numerical approximations can be used. In this letter, we highlight this point by providing an illustrative example of solving for the balancing intercept for several different distributions along with corresponding R and Python code.
DOI: 10.1073/pnas.0307752101
发表时间: 2004-04-06
影响因子: 11.1
作者:
Griffiths, TL;Steyvers, M
通讯作者: Steyvers, M
DOI: --
发表时间: 2009
期刊:
影响因子: --
作者:
Brian Locke
通讯作者: Brian Locke
短信标准化:两个比喻比一个更好吗?
DOI: --
发表时间: 2008
期刊: International Conference on Computational Linguistics
影响因子: --
作者:
Catherine Kobus;François Yvon;Géraldine Damnati
通讯作者: Géraldine Damnati
使用吉布斯采样的 MIMO 信道估计中的权重优化
DOI: --
发表时间: 2021
期刊:
影响因子: --
作者:
Shota Hayakawa;Nobuyuki Hirami;Ibuki Nakamura;and Hisato Fujisaka;宮北和之,佐藤風雅,中野敬介;堀川裕貴・石川博康;佐々木重信・齋藤瑞奈;中馬健士郎 眞田幸俊
通讯作者: 中馬健士郎 眞田幸俊
DOI: --
发表时间: 2010-10
期刊: --
影响因子: --
作者:
Tara McIntosh
通讯作者: Tara McIntosh