R语言如何筛选多列含指定字符串的行并完成受试者分组

方案说明 #

采用tidyverse生态的dplyr+stringr实现，代码简洁易读，适配纵向数据的行级筛选和受试者级分组需求，也提供Base R无依赖备选方案。

前置准备 #

• 安装加载依赖包

# 未安装的话先执行 install.packages("tidyverse")
library(tidyverse)

• 构造示例数据集

df <- tibble(
  id = c(1,1,2,2,2,3,4,4,5),
  dg1 = c("A1","B3","B5","C1","D1","D1","E1","E2","F1"),
  dg2 = c("F1","A13","D1","C1","C1","B4","B1","B7","B3"),
  dg3 = c("D1","C1","B1","B4","B5","D1","G1","F1","D1"),
  dg4 = c("B3","B5","B6","B2","A1","C2","C3","S13","B2"),
  agegroup = c(1,2,1,2,3,1,2,3,1)
)

需求1：多列匹配筛选行 #

首先定义要匹配的目标字符串向量，用if_any()对所有dg开头的列做批量判断，只要任意一列命中就保留该行：

# 定义待匹配的目标字符串
target_str <- c("A1", "A13", "B1", "B2")

# 筛选符合条件的行，保留所有原始列
df_filtered <- df %>%
  filter(if_any(starts_with("dg"), ~ .x %in% target_str))

如果需要模糊匹配（只要包含目标字符串就算命中，不需要完全相等），把判断条件替换为str_detect(.x, paste(target_str, collapse = "|"))即可。

需求2：按受试者（id）划分组别 #

因为是受试者级别的分组，只要该id下存在任意一行符合筛选条件，就划分为A组，否则划分为B组：

df_with_group <- df %>%
  group_by(id) %>%
  mutate(
    has_target = any(if_any(starts_with("dg"), ~ .x %in% target_str)),
    group = ifelse(has_target, "A", "B")
  ) %>%
  ungroup() %>%
  select(-has_target) # 可选，删除中间判断列

执行后group列就是对应的分组结果，示例数据中仅id=3没有命中记录，会被划分为B组，其余id均为A组。

备选：Base R无依赖实现方案 #

如果不想加载第三方包，可以用原生语法实现：

target_str <- c("A1", "A13", "B1", "B2")
# 定位dg列的索引
dg_cols <- grep("^dg", colnames(df))

# 需求1：筛选行
df_filtered <- df[rowSums(sapply(df[dg_cols], function(x) x %in% target_str)) > 0, ]

# 需求2：分组
df$group <- ave(
  rowSums(sapply(df[dg_cols], function(x) x %in% target_str)) > 0,
  df$id,
  FUN = function(x) ifelse(any(x), "A", "B")
)

内容的提问来源于stack exchange，提问作者birch