如何忽略NA值，检查多列是否逐行等价？

需求：检查数据框中指定列的每行内所有非NA值是否相等；若某行全部为NA，则返回NA。

示例数据 #

library(dplyr)
library(tibble)

dat <- tribble(
  ~a, ~b, ~c, ~d,
  4, "blue", "blue", NA,
  7, NA, "red", "green",
  8, "green", "green", "green",
  6, "blue", NA, NA,
  1, NA, NA, NA
)

dat
#> # A tibble: 5 × 4
#>       a b     c     d    
#>   <dbl> <chr> <chr> <chr>
#> 1     4 blue  blue  <NA> 
#> 2     7 <NA>  red   green
#> 3     8 green green green
#> 4     6 blue  <NA>  <NA> 
#> 5     1 <NA>  <NA>  <NA>

期望输出 #

以检查b:d列为例，期望得到如下结果：

expected <- tribble(
  ~a, ~b, ~c, ~d, ~agreement,
  4, "blue", "blue", NA, TRUE,
  7, NA, "red", "green", FALSE,
  8, "green", "green", "green", TRUE,
  6, "blue", NA, NA, TRUE,
  1, NA, NA, NA, NA
)

expected
#> # A tibble: 5 × 5
#>       a b     c     d     agreement
#>   <dbl> <chr> <chr> <chr> <lgl>    
#> 1     4 blue  blue  <NA>  TRUE     
#> 2     7 <NA>  red   green FALSE    
#> 3     8 green green green TRUE     
#> 4     6 blue  <NA>  <NA>  TRUE     
#> 5     1 <NA>  <NA>  <NA>  NA

解决方案 #

方法1：tidyverse风格 #

用dplyr的行处理函数结合条件判断实现：

dat %>%
  rowwise() %>%
  mutate(agreement = case_when(
    all(is.na(c_across(b:d))) ~ NA,
    n_distinct(c_across(b:d), na.rm = TRUE) == 1 ~ TRUE,
    TRUE ~ FALSE
  )) %>%
  ungroup()

方法2：基础R风格 #

自定义判断函数，结合apply逐行处理：

check_agreement <- function(row) {
  non_na_vals <- row[!is.na(row)]
  if (length(non_na_vals) == 0) return(NA)
  all(non_na_vals == non_na_vals[1])
}

dat$agreement <- apply(dat[, c("b", "c", "d")], 1, check_agreement)

两种方法均可得到与期望输出一致的结果。

内容的提问来源于stack exchange，提问作者mac