如何在R语言中按列值分组提取对应关联人名？

需求说明 #

将数据框中a1、a2、a3列里的人名，按对应列的取值（a/b）分组，忽略NA值，最终输出每列的分组结果。

原始数据 #

name <- c("luis", "John", "Leo")
a1 <- c("a","b","a")
a2 <- c("b","a","b")
a3 <- c(NA,"b","a")

df <- data.frame(name,a1,a2,a3)

解决方案 #

方法1：基础R实现 #

遍历目标列，用split按列值分组人名，同时过滤NA对应的无效分组，最后整理成期望格式：

# 遍历a1到a3列
result <- lapply(df[, -1], function(col) {
  # 剔除NA行后，按列值分组对应人名
  split(df$name[!is.na(col)], col[!is.na(col)])
})

# 按要求格式输出结果
for (col_name in names(result)) {
  groups <- result[[col_name]]
  # 把每组人名转成对应字符串格式
  group_strs <- sapply(groups, function(nms) {
    if (length(nms) > 1) {
      paste0("c(\"", paste(nms, collapse = "\",\""), "\")")
    } else {
      paste0("\"", nms, "\"")
    }
  })
  output_str <- paste0(names(group_strs), "=", group_strs, collapse = ", ")
  cat(paste0(col_name, ": ", output_str, "\n"))
}

方法2：tidyverse工具包实现 #

先将数据转为长格式，再分组整理后输出：

library(tidyverse)

df %>%
  pivot_longer(cols = -name, names_to = "col", values_to = "value") %>%
  drop_na(value) %>%
  group_by(col, value) %>%
  summarise(names = list(name), .groups = "drop") %>%
  group_by(col) %>%
  summarise(output = str_c(value, "=", ifelse(map_int(names, length) > 1,
                                             str_c("c(\"", str_c(names, collapse = "\",\""), "\")"),
                                             str_c("\"", names, "\"")),
                           collapse = ", ")) %>%
  mutate(output = str_c(col, ": ", output)) %>%
  pull(output) %>%
  walk(cat, "\n")

运行结果 #

a1: a=c("luis","Leo"), b="John"
a2: a="John", b=c("luis","Leo")
a3: a="Leo", b="John"

内容的提问来源于stack exchange，提问作者Johan