r 替换 lme S4 对象中的部分字符串

Question

这个问题是关于如何从S4 lme模型摘要对象中删除部分字符串。

假设这是我的模型。

# ----preliminaries-------------------------------------------------------
library("lme4")
library("lattice")
library("minqa")
library("nlme")

##-------------------------------------------------------------------------
data("sleepstudy")
##-------------------------------------------------------------------------
sleepstudy$Days[sleepstudy$Days == "0"] <- "Zero"
sleepstudy$Days[sleepstudy$Days == "1"] <- "One"
sleepstudy$Days[sleepstudy$Days == "2"] <- "Two"
sleepstudy$Days[sleepstudy$Days == "3"] <- "Three"
sleepstudy$Days[sleepstudy$Days == "4"] <- "Four"
sleepstudy$Days[sleepstudy$Days == "5"] <- "Five"
sleepstudy$Days[sleepstudy$Days == "6"] <- "Six"
sleepstudy$Days[sleepstudy$Days == "7"] <- "Seven"

## Model
##-------------------------------------------------------------------------
fm1 <- lme(fixed = Reaction ~ Days ,
           random = ~1 | Days, 
           data = sleepstudy)
summary(fm1)

Linear mixed-effects model fit by REML
  Data: sleepstudy 
       AIC      BIC    logLik
  1855.854 1893.483 -915.9268

Random effects:
 Formula: ~1 | Days
        (Intercept) Residual
StdDev:    30.55409 48.61125

Fixed effects:  Reaction ~ Days 
               Value Std.Error  DF   t-value p-value
(Intercept) 336.6295  32.63178 170 10.316003       0
Days9        14.2217  46.14830   0  0.308174     NaN
DaysFive    -28.1110  46.14830   0 -0.609146     NaN
DaysFour    -47.9801  46.14830   0 -1.039693     NaN
DaysOne     -72.1337  46.14830   0 -1.563086     NaN
DaysSeven   -17.8789  46.14830   0 -0.387423     NaN
DaysSix     -24.4512  46.14830   0 -0.529841     NaN
DaysThree   -53.6375  46.14830   0 -1.162285     NaN
DaysTwo     -71.2676  46.14830   0 -1.544317     NaN
DaysZero    -79.9777  46.14830   0 -1.733058     NaN

我喜欢删除fixed effects model summary开头的“

Day

”一词

期望我的固定效应摘要术语看起来像这样。

Fixed effects:  Reaction ~ Days 
               Value Std.Error  DF   t-value p-value
(Intercept) 336.6295  32.63178 170 10.316003       0
9        14.2217  46.14830   0  0.308174     NaN
Five    -28.1110  46.14830   0 -0.609146     NaN
Four    -47.9801  46.14830   0 -1.039693     NaN
One     -72.1337  46.14830   0 -1.563086     NaN
Seven   -17.8789  46.14830   0 -0.387423     NaN
Six     -24.4512  46.14830   0 -0.529841     NaN
Three   -53.6375  46.14830   0 -1.162285     NaN
Two     -71.2676  46.14830   0 -1.544317     NaN
Zero    -79.9777  46.14830   0 -1.733058     NaN

我已经尝试过了

 summary(fm1) <- gsub("Days", "", summary(fm1))

我收到一条错误消息，该对象

fm1

属于 S4 类，gsub 将无法工作。因此，任何关于如何从模型摘要术语中排除“天”一词的建议都是值得赞赏的。谢谢。

Answer 1

您可以使用

broom.mixed::tidy()

将摘要转换为小标题并进行操作：

library(tidyverse)
broom.mixed::tidy(fm1) %>% 
filter(effect == "fixed") %>% # Get fixed effects only
mutate(term = gsub("Days", "", term)) %>%
rename("t-value" = "statistic")

Answer 2

您可以创建自己的函数来模仿

nlme:::summary.lme

函数，如下所示：

在 R 控制台中键入以下内容，并将输出复制到名为“summary.lme.R”的新脚本文件中。

nlme:::summary.lme

通过在第 1 行后插入以下行来编辑此文件：

names(fixed) <- gsub("Days", "", names(fixed))

并在

summary.lme <-

 前面添加 

function(object, adjustSigmma = TRUE, ...)

然后保存文件，将其源到 R 控制台，以便在当前会话中可用，然后再次运行摘要命令。

> summary(fm1)
Linear mixed-effects model fit by REML
  Data: sleepstudy 
       AIC      BIC    logLik
  1855.854 1893.483 -915.9268

Random effects:
 Formula: ~1 | Days
        (Intercept) Residual
StdDev:    30.55409 48.61125

Fixed effects:  Reaction ~ Days 
               Value Std.Error  DF   t-value p-value
(Intercept) 336.6295  32.63178 170 10.316003       0
9            14.2217  46.14830   0  0.308174     NaN
Five        -28.1110  46.14830   0 -0.609146     NaN
Four        -47.9801  46.14830   0 -1.039693     NaN
One         -72.1337  46.14830   0 -1.563086     NaN
Seven       -17.8789  46.14830   0 -0.387423     NaN
Six         -24.4512  46.14830   0 -0.529841     NaN
Three       -53.6375  46.14830   0 -1.162285     NaN
Two         -71.2676  46.14830   0 -1.544317     NaN
Zero        -79.9777  46.14830   0 -1.733058     NaN

您可以使这个函数更通用，但这回答了您的问题。