使用R按预设条件将宽数据框转换为长数据框

首先将你的原始数据框赋值给变量df：

df <- structure(list(study_id = structure(c(1, 3, 4), label = "Subject Study ID"), 
    creatinine1 = structure(c(82, 140, 88), label = "Baseline Creatinine"), 
    creatinine2 = structure(c(90, 122, 125), label = "Creatinie level: 3 months"), 
    egfr1 = structure(c(85, 42, 75), label = "Baseline eGFR "), 
    egfr2 = structure(c(73, 50, 50), label = "eGFR levels : 3 months"), 
    treat = structure(c(1L, 2L, 2L), levels = c("Active", "Placebo"
    ), class = "factor")), row.names = c(NA, -3L), class = c("tbl_df", 
"tbl", "data.frame"))

使用tidyverse工具链完成格式转换：

library(tidyverse)

result_df <- df %>%
  # 将指标列转为长格式，保留分组列
  pivot_longer(cols = -c(study_id, treat), names_to = "name", values_to = "value") %>%
  # 拆分列名：分离生物标志物名称和时间后缀（1/2）
  separate(name, into = c("biomarker", "time"), sep = "(?=\\d)", convert = TRUE) %>%
  # 映射时间后缀为目标列名，同时统一treat列大小写
  mutate(
    time = case_when(
      time == 1 ~ "baseline",
      time == 2 ~ "three_months"
    ),
    treat = tolower(as.character(treat))
  ) %>%
  # 转回宽格式，匹配目标结构
  pivot_wider(names_from = time, values_from = value) %>%
  # 调整生物标志物名称首字母大写
  mutate(biomarker = str_to_title(biomarker)) %>%
  # 调整列顺序与目标一致
  select(study_id, biomarker, baseline, three_months, treat)

# 查看最终结果
print(result_df)

关键步骤说明 #

• pivot_longer：先把宽格式的多指标列转为长格式，便于拆分列名
• separate：通过正则表达式(?=\\d)（正向预查数字）拆分列名，自动区分生物标志物前缀和时间后缀
• case_when：将数字后缀1/2映射为目标列名baseline/three_months
• pivot_wider：将长格式数据转回所需的宽格式，每个生物标志物对应一行数据
• 最后调整文本格式和列顺序，完全匹配目标结构

运行后输出结果：

# A tibble: 6 × 5
  study_id biomarker baseline three_months treat  
     <dbl> <chr>        <dbl>        <dbl> <chr>  
1        1 Creatinine      82           90 active  
2        1 eGFR            85           73 active  
3        3 Creatinine     140          122 placebo 
4        3 eGFR            42           50 placebo 
5        4 Creatinine      88          125 placebo 
6        4 eGFR            75           50 placebo 

内容的提问来源于stack exchange，提问作者kam