R：向表输出添加脚注

Question

我，一个R初学者，创建了以下可重现的示例代码，用于输出各种变量的ADF测试的测试统计信息：

library(urca)
data(nporg)
df <- nporg[complete.cases(nporg), ]

table_adf = function(x) {

  for (i in 1:x) { #something is wrong here
     out1 <- ur.df(x, type = "drift", selectlags = "AIC")@teststat[1]
     out2 <- ur.df(x, type = "trend", selectlags = "AIC")@teststat[1]
     out <- c(out1,out2)
     return(data.frame(matrix(out, nrow=1, dimnames=list(NULL, names(out)))))
  }

  footnote <- rep("", 3)
  footnote[1] <- c("Lag length determined with the AIC information criteria.")
  footnote[2] <- c(paste("ADF critical values drift:", out1@cval[[1]], out1@cval[[2]], out1@cval[[3]]))
  footnote[3] <- c(paste("ADF critical values trend:", out2@cval[[2]], out2@cval[[2]], out2@cval[[3]]))
  print(footnote) # something is wrong here, too
}

rbind(
  table_adf(df$emp),
  table_adf(df$cpi),
  table_adf(df$sp)
)

我的问题： 如何打印表格下方的脚注，如果可能的话，还要将每个变量名称添加到行中？

如果您在我的代码中发现任何其他错误或效率低下，请给我一个提示。 谢谢。

更新：所需输出：

==============================================
                     drift          trend     
----------------------------------------------
emp                 -0.259          -4.841 ***    
cpi                  0.488          -1.326    
sp                   0.232          -3.831 **    
...                  ...            ...
----------------------------------------------
ADF critical values drift: -3.51 -2.89 -2.58  
ADF critical values trend: -4.04 -3.45 -3.15 

我更改了趋势中的数字，以便您可以看到重要性。

Answer 1

每个变量有一个表，有自己的临界值

stargazer具有出色的报告功能。 这是你可以做的事情，例如使用stargazer()的notes=脚注的参数：

library(urca)
library(stargazer)
library(dplyr)

data(nporg)
df <- nporg[complete.cases(nporg), ]

table_adf = function(DF, ...){
  var_list = list(...)
    select_(DF, .dots = var_list) %>%
    {invisible(Map(function(x, y){
      out1 = ur.df(x, type = "drift", selectlags = "AIC")
      out2 = ur.df(x, type = "trend", selectlags = "AIC")
      est_df = data.frame(drift = out1@teststat[1], trend = out2@teststat[1],
                          row.names = y)
      footnote1 = paste("ADF critical values drift:", out1@cval[1,1], out1@cval[1,2], out1@cval[1,3])
      footnote2 = paste("ADF critical values trend:", out2@cval[1,1], out2@cval[1,2], out2@cval[1,3])
      stargazer(est_df, type = "text", summary = FALSE,
                notes = c(footnote1, footnote2))
    },. , var_list))}
}

table_adf(df, "emp", "cpi", "sp")

使用...参数和select_ + .dots而不是select以编程方式提供变量名称

请注意，我已经更改了out1@cval的索引以正确引用“tau”临界值。 虽然我不熟悉ADF测试，但你用out1@cval[[1]]索引2x3矩阵的方式对我来说似乎很奇怪。 所以我怀疑你真的想要那张桌子的第一行。

> out1@cval
      1pct  5pct 10pct
tau2 -3.51 -2.89 -2.58
phi1  6.70  4.71  3.86

结果：

==============================================
                     drift          trend     
----------------------------------------------
emp                 -0.259          -3.041    
----------------------------------------------
ADF critical values drift: -3.51 -2.89 -2.58  
ADF critical values trend: -4.04 -3.45 -3.15  

==============================================
                     drift          trend     
----------------------------------------------
cpi                  0.488          -1.326    
----------------------------------------------
ADF critical values drift: -3.51 -2.89 -2.58  
ADF critical values trend: -4.04 -3.45 -3.15  

==============================================
                     drift          trend     
----------------------------------------------
sp                   0.232          -1.343    
----------------------------------------------
ADF critical values drift: -3.51 -2.89 -2.58  
ADF critical values trend: -4.04 -3.45 -3.15  

单表具有所有变量估计显着性指标

现在，这需要更多参与，因为你必须...：

1. 从每个ur.df对象中提取估计值（ @teststate ）
2. 与@cval相比，查看每个估算值是否显着
3. 将所有估算值放在同一个表格中

注意我在这个函数中使用了最少数量的管道运算符%>% ，因为几乎每个步骤都被使用了不止一次。 举例来说，我不能只是管...参数到lapply的tests ，因为我还需要变量列表设置row.names在est_table一步。

var_list = as.list(unlist(list(...)))似乎不必要地复杂，但它是一种允许用户输入变量名称c("emp", "cpi", "sp")向量的方法c("emp", "cpi", "sp") ，个别变量名称"emp", "cpi", "sp"或两者都是c("emp", "cpi"), "sp" 。 这为用户创造了额外的便利。 最开始的ifelse步骤检查变量名称是否有零参数，如果是这种情况则使用整个数据帧。

table_adf = function(DF, ...){
  # Stores ... arguments as list
  var_list = as.list(unlist(list(...)))

  # Subset DF based on var_list
  if(length(var_list) == 0){
    subset = DF
    var_list = as.list(names(DF))
  }else{
    subset = select_(DF, .dots = var_list)
  }

  # Run model for each variable, and stored as list of lists, each element of the 
  # outer list contains a list of the variable's drift and trend estimates
  tests = lapply(subset, function(x, y){
    out1 = ur.df(x, type = "drift", selectlags = "AIC")
    out2 = ur.df(x, type = "trend", selectlags = "AIC")
    return(list(out1, out2))
  })

  # Store each model estimates in to a dataframe with corresponding variable names
  est_df = lapply(tests, function(x){
    data.frame(drift = x[[1]]@teststat[1], trend = x[[2]]@teststat[1])
  })

  # Combine the list into a single dataframe
  est_table = do.call(rbind, est_df)

  critical_vals = data.frame(tests[[1]][[1]]@cval[1,], 
                             tests[[1]][[2]]@cval[1,])

  # Check for significance level and add *'s accordingly
  est_table = Map(function(x, y){
    ifelse(x < y[1], paste(x, "***"), 
           ifelse(x < y[2], paste(x, "**"),
                  ifelse(x < y[3], paste(x, "*"), x)))
  }, est_table, critical_vals) %>% data.frame(row.names = unlist(var_list))

  # Set footnotes
  footnote1 = paste("ADF critical values drift:", paste(critical_vals[[1]], collapse = " "))
  footnote2 = paste("ADF critical values trend:", paste(critical_vals[[2]], collapse = " "))

  # Create nice looking table with critical values  
  stargazer(est_table, type = "text", summary = FALSE,
            notes = c(footnote1, footnote2))
}

下面演示了用户输入变量名称以获得所需估计值的方法：

> table_adf(df, "emp", "cpi", "sp")

==============================================
                     drift          trend     
----------------------------------------------
emp                 -0.259          -3.041    
cpi                  0.488          -1.326    
sp                   0.232          -1.343    
----------------------------------------------
ADF critical values drift: -3.51 -2.89 -2.58  
ADF critical values trend: -4.04 -3.45 -3.15  

> table_adf(df, names(df)[-1])

==============================================
                     drift           trend    
----------------------------------------------
gnp.r          1.22473721812295      -1.308   
gnp.n           4.2258196600637      2.195    
gnp.pc        -0.153914521574277     -2.628   
ip             1.49350460242673      -0.924   
emp           -0.258523412908067     -3.041   
ur            -2.75061777744267 *    -2.777   
gnp.p          1.06909915298064      -0.800   
cpi            0.488243115715459     -1.326   
wg.n           2.81042091873464      0.399    
wg.r           0.399440467221402     -2.445   
M              1.45354783780152      -0.454   
vel            -2.50304919304012     -2.537   
bnd            1.21565390985417      1.837    
sp             0.231913401411228     -1.343   
----------------------------------------------
ADF critical values drift: -3.51 -2.89 -2.58  
ADF critical values trend: -4.04 -3.45 -3.15  

> table_adf(df)

==============================================
                     drift           trend    
----------------------------------------------
year           -1.73205080756904     -1.732   
gnp.r          1.22473721812295      -1.308   
gnp.n           4.2258196600637      2.195    
gnp.pc        -0.153914521574277     -2.628   
ip             1.49350460242673      -0.924   
emp           -0.258523412908067     -3.041   
ur            -2.75061777744267 *    -2.777   
gnp.p          1.06909915298064      -0.800   
cpi            0.488243115715459     -1.326   
wg.n           2.81042091873464      0.399    
wg.r           0.399440467221402     -2.445   
M              1.45354783780152      -0.454   
vel            -2.50304919304012     -2.537   
bnd            1.21565390985417      1.837    
sp             0.231913401411228     -1.343   
----------------------------------------------
ADF critical values drift: -3.51 -2.89 -2.58  
ADF critical values trend: -4.04 -3.45 -3.15