如何在R中对非完全重复的配对型记录进行去重？

你遇到的这个问题其实很典型——这是无序配对组合的去重场景：比如(2,43)和(43,2)本质是同一个配对，但因为两列顺序相反，常规的unique()或者dplyr::distinct()这类去重函数没法识别它们是重复项。下面给你两种实用的解决方案，不管用base R还是tidyverse工具都能搞定：

方法一：用dplyr（tidyverse风格） #

我们可以给每一行生成一个「标准化配对键」——把A和B的元素按固定顺序排序后拼接成字符串，这样所有本质相同的配对都会生成同一个键，再基于这个键去重就简单了：

library(dplyr)

# 你的原始数据
df <- data.frame( 
  A = c("2","2","2","43","43","43","331","391","481","490","501","501","501","502","502","502"), 
  B = c("43","501","502","2","501","502","491","496","490","481","2","43","502","2","43","501")
)

# 生成标准化键并去重
df_unique <- df %>%
  rowwise() %>%
  # 把每行的A、B排序后拼接成唯一键
  mutate(pair_key = paste(sort(c(A, B)), collapse = "-")) %>%
  ungroup() %>%
  # 保留每个唯一键对应的第一行
  distinct(pair_key, .keep_all = TRUE) %>%
  # 删掉临时生成的键列
  select(-pair_key)

# 查看结果
print(df_unique)

运行后你会得到去重后的结果，所有像(2,43)和(43,2)这类重复配对只会保留一条。

方法二：用base R（无需额外包） #

如果习惯用原生R代码，也可以直接通过按行排序+去重的方式实现：

# 你的原始数据
df <- data.frame( 
  A = c("2","2","2","43","43","43","331","391","481","490","501","501","501","502","502","502"), 
  B = c("43","501","502","2","501","502","491","496","490","481","2","43","502","2","43","501")
)

# 按行排序后，用duplicated识别重复项，提取非重复行
df_unique <- df[!duplicated(t(apply(df, 1, sort))), ]

# 查看结果
print(df_unique)

这个方法更简洁：apply(df,1,sort)按行对A、B排序，t()转置后用duplicated()找出重复的配对，最后用!取反得到非重复行的索引，提取即可。

两种方法最终都会得到你想要的目标输出：

A    B
1    2   43
2    2  501
3    2  502
5   43  501
6   43  502
7  331  491
8  391  496
9  481  490

内容的提问来源于stack exchange，提问作者user3357059