R：使用转换表更新 data.frame 中的行名

Question

我有一个如下形式的data.frame：

> set.seed(1)
> myp <- paste0('P', sort(sample(1:15, 10)))
> mydf <- data.frame(param=c(paste0(myp, 'B'), paste0(myp, 'R'), paste0(myp, 'max'), paste0(myp, 'min')),
+ value1=NA, value2=NA, value3=NA)
> rownames(mydf) <- mydf$param
> mydf$param <- NULL
> mydf
       value1 value2 value3
P1B        NA     NA     NA
P3B        NA     NA     NA
P4B        NA     NA     NA
P5B        NA     NA     NA
P6B        NA     NA     NA
P8B        NA     NA     NA
P9B        NA     NA     NA
P10B       NA     NA     NA
P11B       NA     NA     NA
P14B       NA     NA     NA
P1R        NA     NA     NA
P3R        NA     NA     NA
P4R        NA     NA     NA
P5R        NA     NA     NA
P6R        NA     NA     NA
P8R        NA     NA     NA
P9R        NA     NA     NA
P10R       NA     NA     NA
P11R       NA     NA     NA
P14R       NA     NA     NA
P1max      NA     NA     NA
P3max      NA     NA     NA
P4max      NA     NA     NA
P5max      NA     NA     NA
P6max      NA     NA     NA
P8max      NA     NA     NA
P9max      NA     NA     NA
P10max     NA     NA     NA
P11max     NA     NA     NA
P14max     NA     NA     NA
P1min      NA     NA     NA
P3min      NA     NA     NA
P4min      NA     NA     NA
P5min      NA     NA     NA
P6min      NA     NA     NA
P8min      NA     NA     NA
P9min      NA     NA     NA
P10min     NA     NA     NA
P11min     NA     NA     NA
P14min     NA     NA     NA

我想更新行的名称，为此我有一个如下所示的转换表：

> conv.df <- data.frame(name=myp, new.name=paste0('P', 1:10))
> conv.df
   name new.name
1    P1       P1
2    P3       P2
3    P4       P3
4    P5       P4
5    P6       P5
6    P8       P6
7    P9       P7
8   P10       P8
9   P11       P9
10  P14      P10

我唯一想做的就是更新mydf 中的行名，使其反映转换表conv.df 中的new.names。

这看起来很简单，但我无法理解它...我会尝试regular expression 方法，我唯一的问题是我不能 100% 确定我只会遇到rownames 形式的@ 987654330@、PnR、Pnmax、Pnmin... 我想要一个适用于任何PnX 实例的解决方案（它始终是Pn，后跟[:alpha:]）

Answer 1

您可以将行名设为一列，将根“Pn”和字母“[:alpha:]”拆分，然后将它们重命名如下，

set.seed(1)
myp <- paste0('P', sort(sample(1:15, 10)))
mydf <- data.frame(param=c(paste0(myp, 'B'), paste0(myp, 'R'), paste0(myp, 'max'), paste0(myp, 'min')),
                     value1=NA, value2=NA, value3=NA)
rownames(mydf) <- mydf$param
mydf$param <- NULL
mydf


library(tidyverse)

mydf%>%
  rownames_to_column()%>%
  mutate(root = gsub("^(P\\d\\d?).*$","\\1",rowname),
         letter = gsub("^P\\d\\d?(.*)$","\\1",rowname))%>%
  mutate(root = recode(root,
    P1 = "P1",
    P3 = "P2",
    P4 = "P3",
    P5 = "P4",
    P6 = "P5",
    P8 = "P6",
    P9 = "P7",
    P10 = "P8",
    P11 = "P9",
    P14 = "P10"
  ))%>%
  mutate(rowname = paste0(root,letter))%>%
  column_to_rownames()%>%
  select(-root,-letter)

@teofil 的回答很好，而且有效。这是另一种仍然使用重新编码和一些元编程的方法，

library(tidyverse)

rename_col_df <- function(data,colname,df_rename){
  # data is the input data frame
  # colname is the column to be modified
  # df_rename must have columns name and new.name

  colname = enexpr(colname) # Capture the user input col name as a symbol
  old_name = df_rename$name
  new_name = df_rename$new.name

  # Start construcing an expression
  # The following line creates a recode function
  # recode_expr[[1]] is "recode"
  # recode_expr[[2]] is the first argument
  recode_expr = expr(recode(!!colname))

  # All subsequent arguments to recode are added here

  for(i in seq_along(old_name)){
    recode_expr[[old_name[i]]] = new_name[i]
  }


  data = data %>% mutate(!!colname := !!recode_expr)

  return(data)

}

conv.df <- data.frame(name=myp, new.name=paste0('P', 1:10),stringsAsFactors = FALSE)

mydf%>%
  rownames_to_column()%>%
  mutate(root = gsub("^(P\\d\\d?).*$","\\1",rowname),
         letter = gsub("^P\\d\\d?(.*)$","\\1",rowname))%>%
  rename_col_df(root,conv.df) %>%
  mutate(rowname = paste0(root,letter))%>%
  column_to_rownames()%>%
  select(-root,-letter)

如果我们想将重命名应用于向量而不是数据框，

rename_vec_df <- function(vec,df_rename){
  # vec is the vector to be modified
  # df_rename must have columns name and new.name

  old_name = df_rename$name
  new_name = df_rename$new.name

  # Start construcing an expression
  # The following line creates a recode function
  # recode_expr[[1]] is "recode"
  # recode_expr[[2]] is the first argument
  recode_expr = expr(recode(!!vec))

  # All subsequent arguments to recode are added here

  for(i in seq_along(old_name)){
    recode_expr[[old_name[i]]] = new_name[i]
  }


  vec = eval(recode_expr)

  return(vec)

}

myp <- paste0('P', sort(sample(1:15, 10)))
conv.df <- data.frame(name=myp, new.name=paste0('P', 1:10),stringsAsFactors = FALSE)

vec <- c("P1","P14","P10")

rename_vec_df(vec,conv.df)

要详细了解此处使用的技术，

https://rlang.r-lib.org/reference/quotation.html https://adv-r.hadley.nz/metaprogramming.html

Answer 2

数据框中不能有重复的行名。这是基本 R 中将其添加为列名的一种方法。这里我们提取原始rownames中的共同部分，即“P”后跟一个数字，match它与conv.df$name并得到对应的conv.df$new.name。

mydf$new_name <- conv.df$new.name[
                match(sub("(P\\d+).*", "\\1", rownames(mydf)), conv.df$name)]


mydf
#       value1 value2 value3 new_name
#P1B        NA     NA     NA       P1
#P2B        NA     NA     NA       P2
#P3B        NA     NA     NA       P3
#P4B        NA     NA     NA       P4
#P7B        NA     NA     NA       P5
#P8B        NA     NA     NA       P6
#P9B        NA     NA     NA       P7
#P11B       NA     NA     NA       P8
#P12B       NA     NA     NA       P9
#P13B       NA     NA     NA      P10
#P1R        NA     NA     NA       P1
#P2R        NA     NA     NA       P2
#...

在哪里

sub("(P\\d+).*", "\\1", rownames(mydf)) #returns

#[1] "P1"  "P2"  "P3"  "P4"  "P7"  "P8"  "P9"  "P11" "P12" "P13" "P1"  "P2"  "P3"  
#    "P4"  "P7"  "P8"  "P9"  "P11" "P12" "P13" "P1"  "P2"  "P3"  "P4"  "P7"  "P8"  
#    "P9"  "P11" "P12" "P13" "P1"  "P2"  "P3"  "P4"  "P7"  "P8"  "P9"  "P11" "P12" 
#    "P13"

Answer 3

跟进@Sada93 代码。要避免recode 步骤，请使用join：

library(tidyverse)
set.seed(1)
myp <- paste0('P', sort(sample(1:15, 10)))
mydf <-
  data.frame(
    param = c(
      paste0(myp, 'B'),
      paste0(myp, 'R'),
      paste0(myp, 'max'),
      paste0(myp, 'min')
    ),
    value1 = NA,
    value2 = NA,
    value3 = NA
  )
rownames(mydf) <- mydf$param
mydf$param <- NULL
mydf

conv.df <- data.frame(name=myp, new.name=paste0('P', 1:10))

mydf %>% rownames_to_column() %>%
  mutate(name = gsub("^(P\\d\\d?).*$","\\1",rowname),
         letter = gsub("^P\\d\\d?(.*)$","\\1",rowname)) %>% 
  left_join(., conv.df, by="name") %>% 
  mutate(rowname=paste(new.name, letter, sep="")) %>% 
  column_to_rownames() %>% 
  select(-name, -letter)

       value1 value2 value3 new.name
P1B        NA     NA     NA       P1
P2B        NA     NA     NA       P2
P3B        NA     NA     NA       P3
P4B        NA     NA     NA       P4
P5B        NA     NA     NA       P5
P6B        NA     NA     NA       P6
P7B        NA     NA     NA       P7
P8B        NA     NA     NA       P8
P9B        NA     NA     NA       P9
P10B       NA     NA     NA      P10
P1R        NA     NA     NA       P1