如何在平行列中向前查找下一个匹配项（R语言tidyverse）

架构老林

2026-6-20

在tidyverse中实现"按to值向下匹配from对应type"的简便方法

需求说明

给数据框新增want列：对每一行，根据当前行to列的值，在from列中向下查找第一个出现该值的行，将该行的type值作为当前行的want；若后续无匹配项，want设为NA。

示例数据

示例1（含预期结果）

library(tidyverse)

d <- tribble(~from, ~to, ~type, ~want,
         "a", "b", 9, 8,
         "b", "c", 8, 0,
         "a", "x", 6, 7,
         "c", "d", 0, 3,
         "c", "d", 5, 3,
         "x", "y", 7, NA,
         "d", "e", 3, 4,
         "e", "f", 4, 5, 
         "e", "f", 2, 5,
         "e", "f", 8, 5,
         "f", "g", 5, NA)

示例2

d2 <- tribble(~from, ~to, ~type,
          "a", "b", 0,
          "b", "a", 1,
          "a", "c", 2)

简便实现方法

用dplyr搭配purrr::map_dbl就能快速搞定，核心思路是对每行的to值，从当前行的下一行开始找from匹配的记录，取第一个对应的type值：

# 处理示例1
d_result <- d %>%
  mutate(want_calc = map_dbl(to, function(target) {
    # 获取当前行之后的所有行数据
    after_rows <- slice(., (row_number() + 1):n())
    # 筛选from等于target的行，取第一个type
    first_match <- after_rows$type[after_rows$from == target][1]
    # 无匹配则返回NA
    ifelse(is.null(first_match), NA_real_, first_match)
  }))

# 验证计算结果与预期是否一致
all.equal(d_result$want_calc, d$want)
# [1] TRUE

处理示例2

直接复用相同逻辑：

d2_result <- d2 %>%
  mutate(want = map_dbl(to, function(target) {
    after_rows <- slice(., (row_number() + 1):n())
    first_match <- after_rows$type[after_rows$from == target][1]
    ifelse(is.null(first_match), NA_real_, first_match)
  }))

# 输出结果
d2_result
# # A tibble: 3 × 4
#   from  to    type  want
#   <chr> <chr> <dbl> <dbl>
# 1 a     b         0     1
# 2 b     a         1     2
# 3 a     c         2    NA