coxRegressionResiduals(WGCNA)
coxRegressionResiduals()所属R语言包：WGCNA
Deviance- and martingale residuals from a Cox regression model
Cox回归模型的越轨行为和鞅残差
译者：生物统计家园网 机器人LoveR
描述----------Description----------
The function inputs a censored time variable which is specified by two input variables time and event. It outputs i) the martingale residual and ii) deviance residual corresponding to a Cox regression model. By default, the Cox regression model is an intercept only Cox regression model. But optionally, the user can input covariates using the argument datCovariates. The function makes use of the coxph function in the survival library. See help(residuals.coxph) to learn more.
该函数的输入变量，它是通过两个输入变量time和event审查的时间。输出I）的鞅剩余的及ii）偏差残余相应的Cox回归模型。默认情况下，Cox回归模型是仅截距Cox回归模型。但是，任选地，用户可以输入协变量使用的参数datCovariates。该函数使得使用的coxph函数库中的生存。请参阅help(residuals.coxph)学习更多。
用法----------Usage----------
coxRegressionResiduals(time, event, datCovariates = NULL)
参数----------Arguments----------
参数：time
is a numeric variable that contains follow up time or time to event.
是一个数字变量，它包含随访时间或事件的时间。
参数：event
is a binary variable that takes on values 1 and 0. 1 means that the event took place (e.g. person died, or tumor recurred). 0 means censored, i.e. event has not yet been observed or loss to follow up.
是一个二进制变量，需要对值1和0。 1表示事件发生（如人死亡，或肿瘤复发）。 0审查，即事件尚未被观察到或亏损跟进。
参数：datCovariates
a data frame whose columns correspond to covariates that should be used in the Cox regression model. By default, the only covariate the intercept term 1.
一个数据框的列相对应的协变量Cox回归模型中，应使用。默认情况下，只有协截距项1。
Details
详细信息----------Details----------
Residuals are often used to investigate the lack of fit of a model. For Cox regression, there is no easy analog to the usual "observed minus predicted" residual of linear regression. Instead, several specialized residuals have been proposed for Cox regression analysis. The function calculates residuals that are well defined for an intercept only Cox regression model: the martingale and deviance residuals (Therneau et al 1990). The martingale residual of a subject (person) specifies excess failures beyond the expected baseline hazard. For example, a subject who was censored at 3 years, and whose predicted cumulative hazard at 3 years was 30 Another subject who had an event at 10 years, and whose predicted cumulative hazard at 10 years was 60 Since martingale residuals are not symmetrically distributed, even when the fitted model is correct, it is often advantageous to transform them into more symmetrically distributed residuals: deviance residuals. Thus, deviance residuals are defined as transformations of the martingale residual and the event variable. Deviance residuals are often symmetrically distributed around zero Deviance Residuals are similar to residuals from ordinary linear regression in that they are symmetrically distributed around 0 and have standard deviation of 1.0. . A subjects with a large deviance residual is poorly predicted by the model, i.e. is different from the baseline cumulative hazard. A negative value indicates a longer than expected survival time. When covariates are specified in datCovariates, then one can plot deviance (or martingale) residuals against the covariates. Unusual patterns may indicate poor fit of the Cox model. Cryptic comments: Deviance (or martingale) residuals can sometimes be used as (uncensored) quantitative variables instead of the original time censored variable. For example, they could be used as outcome in a regression tree or regression forest predictor.
残差经常被用来调查缺乏适合的模型。 Cox回归，有没有简单的模拟通常的“观察减去预测的线性回归的残差。相反，一些专门的残留物已经被提出了Cox回归分析。该函数计算的残差定义的拦截只Cox回归模型：鞅和越轨行为的残差（Therneau等人，1990年）。鞅残余的对象（人）指定了超出预期的基线风险多余的故障。例如，一个受人被检察处3年，而其预计在3年的累积危险为30的另一个问题，有一个事件在10年，并预计在10年的累积危险是60年代以来鞅残差对称分布，即使在合适的模型是正确的，它往往是有利的，把它们转化为对称分布残差偏差残差。因此，偏差残差定义为鞅残留和事件变量的转换。越轨残差往往周围零越轨残差的对称分布是相似的从普通的线性回归的残差，因为它们是对称地分布在0附近，有标准偏差为1.0。 。具有大越轨残余阿科目很差由模型预测的，即是从基线累积风险不同。负值表示比预期更长的生存时间。当协变量中指定的datCovariates，然后可以绘制出偏差（鞅）残差对协变量。不寻常的图案可能表明Cox比例风险模型的拟合差。神秘的评论：的越轨（或鞅）残留物有时可以量化的变量，而不是原来的定时截尾变量（无码）。例如，它们可以被用来作为回归树或回归森林的预测结果。
值----------Value----------
It outputs a data frame with 2 columns. The first and second column correspond to martingale and deviance residuals respectively.
它输出的数据框，2列。第一和第二列对应的鞅和偏差残差分别。
注意----------Note----------
This function can be considered a wrapper of the coxph function.
这个功能可以被认为是一个包装的coxph功能。
（作者）----------Author(s)----------
Steve Horvath
参考文献----------References----------
实例----------Examples----------
library(survival)
# simulate time and event data[模拟时间和事件数据]
time1=sample(1:100)
event1=sample(c(1,0), 100,replace=TRUE)
event1[1:5]=NA
time1[1:5]=NA
# no covariates[没有协变量]
datResiduals= coxRegressionResiduals(time=time1,event=event1)
# now we simulate a covariate[现在我们模拟了协]
z= rnorm(100)
cor(datResiduals,use="p")
datResiduals=coxRegressionResiduals(time=time1,event=event1,datCovariates=data.frame(z))
cor(datResiduals,use="p")
转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。
注：
注1：为了方便大家学习，本文档为生物统计家园网机器人LoveR翻译而成，仅供个人R语言学习参考使用，生物统计家园保留版权。
注2：由于是机器人自动翻译，难免有不准确之处，使用时仔细对照中、英文内容进行反复理解，可以帮助R语言的学习。
注3：如遇到不准确之处，请在本贴的后面进行回帖，我们会逐渐进行修订。
|
发表于 2012-10-1 21:12:11
