r - 同时修剪多个变量的因子标签

Question

这是我的玩具数据集：

 library(tidyverse)
dat <- tibble (x1 = c("False - very long label specific to x1", "False - very long label specific to x1", "True - very long label specific to x1", "True - very long label specific to x1"),
               x2 = c("False - very long label specific to x2", "False - very long label specific to x2", "False - very long label specific to x2", "True - very long label specific to x2"),
               y = c(10, 5, 12, 4)) %>% mutate_at(vars(x1:x2), factor)
head(dat)
#> # A tibble: 4 x 3
#>   x1                                x2                                    y
#>   <fct>                             <fct>                             <dbl>
#> 1 False - very long label specific~ False - very long label specific~    10
#> 2 False - very long label specific~ False - very long label specific~     5
#> 3 True - very long label specific ~ False - very long label specific~    12
#> 4 True - very long label specific ~ True - very long label specific ~     4

我想修剪很长的因子标签，它们都有两个共同点：

1. 都以True或False开头
2. 包括列名（即每列的因子标签是唯一的）

我想简化这一点，并且每个因子列只有True和False 之类的东西。这是我想要的输出：

#> # A tibble: 4 x 3
#>   x1    x2        y
#>   <fct> <fct> <dbl>
#> 1 False False    10
#> 2 False False     5
#> 3 True  False    12
#> 4 True  True      4

我认为它应该与类似mutate_atandfct_relabel和可能的东西一起使用str_trunc，但我无法弄清楚。

Answer 1

我们可以trimws使用whitespace

library(dplyr)
dat %>% 
    mutate_if(is.factor, ~ factor(trimws(., whitespace = "\\s*-.*")))
# A tibble: 4 x 3
#  x1    x2        y
#  <fct> <fct> <dbl>
#1 False False    10
#2 False False     5
#3 True  False    12
#4 True  True      4

---

或fct_relabel与str_remove

library(forcats)
library(stringr)
dat %>% 
    mutate_if(is.factor, ~ fct_relabel(., ~str_remove(., '\\s*-.*')))

---

或使用data.table

library(data.table)
m1 <- names(which(sapply(dat, is.factor)))
setDT(dat)[, (nm1) := lapply(.SD, function(x) 
        factor(sub('\\s*-.*', "", x))) , .SDcols = nm1]