table 中的堆叠回归？

Question

是否有任何 R 包可以 (a) 让您将模型放在相邻的列中，以及 (b) 在不同的行中堆叠回归？我看到了一种可能的解决方案 (R stargazer stack several regression models in rows under each other)，但这并不能完全解决问题，因为它包含所有模型系数并添加了其他垃圾。

下面是我尝试创建的示例。行是因变量，列是模型规格：

我可以通过为单个因变量估计第 1、2、3 列中的模型来创建我想要的一行：

data(mtcars)
attach(mtcars)
library(stargazer)

m1 <- lm(mpg ~ cyl, mtcars)
m2 <- lm(mpg ~ cyl + disp, mtcars)
m3 <- lm(mpg ~ cyl + disp + hp, mtcars)

stargazer(m1,m2,m3, keep = c('cyl'), type = 'text')

Answer 1

这个怎么样。我写了一个小函数来获取模型列表并生成所需的输出。

myfun <- function(..., param_num = 2, cn = NULL, rn=NULL, output=c("text", "md", "latex")){
  ### ... is a list where each element is itself a list of models that
  ###     will be used in in the corresponding column
  ### param_num is a vector of number isdentify the index of the 
  ###     coefficient to be used for that column
  ### cn is a vector of column names for the table. 
  ### rn is a vector of names of the rows of the table
  ### output a string indicating the kind of output you want.  If "md" (markdown), then 
  ###     the knitr package will be loaded.  If "latex", then the xtable package will be loaded.
  ### This function requires the psre package
  require(psre)
  output = match.arg(output)
  args <- list(...)
  if(length(param_num) == 1)param_num <- rep(param_num, length(args))
  b <- lapply(1:length(args), function(x)sapply(args[[x]], function(y)coef(y)[param_num[x]]))
  se <- lapply(1:length(args), function(x)sapply(args[[x]], function(y)sqrt(diag(vcov(y)))[param_num[x]]))
  res_df <- lapply(1:length(args), function(x)sapply(args[[x]], function(y)y$df.residual))
  t <- lapply(1:length(b), function(i)b[[i]]/se[[i]])
  p <- lapply(1:length(b), function(i)2*pt(abs(t[[i]]), df=res_df[[i]], lower.tail=FALSE))
  cols <- sapply(1:length(b), function(i)shuffle(b[[i]], p[[i]], se[[i]]))
  cols <- cbind("", cols)
  if(is.null(cn)){
    colnames(cols) <- c("DV", paste0("Spec ", 1:length(b)))
  }else{
    colnames(cols) <- cn
  }
  if(is.null(rn)){
    cols[seq(1, nrow(cols), by=2), 1] <- paste0("Var ", seq_along(seq(1, nrow(cols), by=2)))
  }else{
    cols[seq(1, nrow(cols), by=2), 1] <- rn
  }
  if(output == "text"){
    print(noquote(cols))
  }
  if(output == "md"){
    require(knitr)
    print(knitr::kable(cols))
  }
  if(output == "latex"){
    require(xtable)
    print(xtable(cols), include.rownames=FALSE)
  }
}


library(mtcars)
#> Error in library(mtcars): there is no package called 'mtcars'
col1_mods <- list(lm(mpg ~ vs + cyl, data=mtcars), 
                  lm(disp ~ vs + cyl, data=mtcars),
                  lm(hp ~ vs + cyl, data=mtcars), 
                  lm(qsec ~ vs + cyl, data=mtcars))
col2_mods <- list(lm(mpg ~ vs , data=mtcars), 
                  lm(disp ~ vs , data=mtcars),
                  lm(hp ~ vs , data=mtcars), 
                  lm(qsec ~ vs , data=mtcars))

col3_mods <- list(lm(mpg ~ vs + cyl + am, data=mtcars), 
                  lm(disp ~ vs + cyl + am, data=mtcars),
                  lm(hp ~ vs + cyl + am, data=mtcars), 
                  lm(qsec ~ vs + cyl + am, data=mtcars))


myfun(col1_mods, col2_mods, col3_mods, 
      rn = c("MPG", "Displacement", "Horsepower", "Time to 60"), 
      output = "text")
#> Loading required package: psre
#>      DV           Spec 1   Spec 2    Spec 3  
#> [1,] MPG          -0.939   7.940*    1.295   
#> [2,]              (1.978)  (1.632)   (2.193) 
#> [3,] Displacement 15.047   -174.693* -20.554 
#> [4,]              (33.561) (31.597)  (37.519)
#> [5,] Horsepower   -19.115  -98.365*  10.364  
#> [6,]              (23.649) (17.155)  (25.813)
#> [7,] Time to 60   2.744*   2.640*    1.063   
#> [8,]              (0.751)  (0.432)   (0.637)


myfun(col1_mods, col2_mods, col3_mods, 
      rn = c("MPG", "Displacement", "Horsepower", "Time to 60"), 
      output = "md")
#> Loading required package: knitr
#> 
#> 
#> |DV           |Spec 1   |Spec 2    |Spec 3   |
#> |:------------|:--------|:---------|:--------|
#> |MPG          |-0.939   |7.940*    |1.295    |
#> |             |(1.978)  |(1.632)   |(2.193)  |
#> |Displacement |15.047   |-174.693* |-20.554  |
#> |             |(33.561) |(31.597)  |(37.519) |
#> |Horsepower   |-19.115  |-98.365*  |10.364   |
#> |             |(23.649) |(17.155)  |(25.813) |
#> |Time to 60   |2.744*   |2.640*    |1.063    |
#> |             |(0.751)  |(0.432)   |(0.637)  |


myfun(col1_mods, col2_mods, col3_mods, 
      rn = c("MPG", "Displacement", "Horsepower", "Time to 60"), 
      output = "latex")
#> Loading required package: xtable
#> 
#> Attaching package: 'xtable'
#> The following object is masked from 'package:psre':
#> 
#>     caption
#> % latex table generated in R 4.1.2 by xtable 1.8-4 package
#> % Fri Feb  4 09:46:07 2022
#> \begin{table}[ht]
#> \centering
#> \begin{tabular}{llll}
#>   \hline
#> DV & Spec 1 & Spec 2 & Spec 3 \ 
#>   \hline
#> MPG & -0.939  & 7.940* & 1.295  \ 
#>    & (1.978) & (1.632) & (2.193) \ 
#>   Displacement & 15.047  & -174.693* & -20.554  \ 
#>    & (33.561) & (31.597) & (37.519) \ 
#>   Horsepower & -19.115  & -98.365* & 10.364  \ 
#>    & (23.649) & (17.155) & (25.813) \ 
#>   Time to 60 & 2.744* & 2.640* & 1.063  \ 
#>    & (0.751) & (0.432) & (0.637) \ 
#>    \hline
#> \end{tabular}
#> \end{table}

^{由 reprex package (v2.0.1)}

于 2022-02-04 创建

table 中的堆叠回归？

Stacking regressions in a table?

r

stargazer