如何优化R语言中基于多列生成Purpose列的代码？

你的嵌套ifelse代码虽然可行，但确实存在扩展性差的问题——新增列时需要修改多层嵌套，代码冗余且易出错。下面先修正你循环代码的问题，再提供几种更优、更易扩展的实现方法。

一、修正你的循环代码 #

你的循环代码存在几个关键错误：

• for (1:1000) 语法错误，需指定循环变量（比如行索引）
• 赋值时索引逻辑错误，Purpose[i] 中的i是列名而非行索引
• 未处理列名到友好名称的转换（如NEW_CAR转New_Car）
• 未实现“找到第一个Yes即停止”的逻辑，也没处理无Yes时的None赋值

修正后的循环代码：

# 定义列名与友好名称的映射表，新增/删除列只需修改这里
purpose_map <- c(
  NEW_CAR = "New_Car",
  USED_CAR = "Used_Car",
  FURNITURE = "Furniture",
  `RADIO/TV` = "Radio/TV",
  EDUCATION = "Education",
  RETRAINING = "Retraining"
)
target_cols <- names(purpose_map)

# 初始化结果列
CD$PURPOSE <- rep("None", nrow(CD))

# 逐行遍历，找到第一个"Yes"的列并赋值对应名称
for (row_idx in 1:nrow(CD)) {
  for (col_name in target_cols) {
    if (CD[row_idx, col_name] == "Yes") {
      CD$PURPOSE[row_idx] <- purpose_map[col_name]
      break  # 找到第一个匹配项后终止当前行的列循环
    }
  }
}

二、Base R 简洁实现（apply） #

用apply按行处理，避免显式循环，代码更紧凑：

purpose_map <- c(
  NEW_CAR = "New_Car",
  USED_CAR = "Used_Car",
  FURNITURE = "Furniture",
  `RADIO/TV` = "Radio/TV",
  EDUCATION = "Education",
  RETRAINING = "Retraining"
)
target_cols <- names(purpose_map)

# 按行提取第一个"Yes"对应的友好名称，无匹配则返回"None"
CD$PURPOSE <- apply(CD[, target_cols], 1, function(row_vals) {
  yes_pos <- which(row_vals == "Yes")[1]  # 取第一个"Yes"的位置
  if (length(yes_pos) == 0) "None" else purpose_map[yes_pos]
})

三、Tidyverse 风格实现（dplyr + tidyr） #

适合习惯tidyverse生态的用户，逻辑清晰，扩展性极强：

library(dplyr)
library(tidyr)

# 定义列名-友好名称映射表，新增列只需添加行
purpose_map <- tibble(
  col_name = c("NEW_CAR", "USED_CAR", "FURNITURE", "RADIO/TV", "EDUCATION", "RETRAINING"),
  purpose = c("New_Car", "Used_Car", "Furniture", "Radio/TV", "Education", "Retraining")
)

CD <- CD %>%
  mutate(row_id = row_number()) %>%  # 添加行号用于后续合并
  pivot_longer(
    cols = all_of(purpose_map$col_name),
    names_to = "col_name",
    values_to = "value"
  ) %>%
  filter(value == "Yes") %>%
  left_join(purpose_map, by = "col_name") %>%
  group_by(row_id) %>%
  slice(1) %>%  # 保留每行第一个"Yes"的记录
  ungroup() %>%
  right_join(CD %>% mutate(row_id = row_number()), by = "row_id") %>%
  mutate(PURPOSE = coalesce(purpose, "None")) %>%  # 无匹配时填充"None"
  select(-row_id, -col_name, -value, -purpose)  # 清理临时列

四、Data.table 高效实现（适合大数据） #

如果你的数据量较大，data.table的实现速度更快，内存效率更高：

library(data.table)

setDT(CD)

purpose_map <- data.table(
  col_name = c("NEW_CAR", "USED_CAR", "FURNITURE", "RADIO/TV", "EDUCATION", "RETRAINING"),
  purpose = c("New_Car", "Used_Car", "Furniture", "Radio/TV", "Education", "Retraining")
)

# 添加行号标识
CD[, row_id := .I]

# 转长格式、筛选"Yes"记录、保留每行第一个匹配项
long_CD <- melt(
  CD,
  id.vars = "row_id",
  measure.vars = purpose_map$col_name,
  variable.name = "col_name",
  value.name = "value"
)[value == "Yes", ][order(row_id), .SD[1], by = row_id]

# 合并映射表与原数据，填充None
long_CD <- long_CD[purpose_map, on = "col_name"]
CD[long_CD, PURPOSE := purpose, on = "row_id"]
CD[is.na(PURPOSE), PURPOSE := "None"]
CD[, row_id := NULL]  # 清理临时列

扩展性说明 #

所有上述方案的核心扩展点都是**purpose_map映射表**：

• 新增列时，只需在映射表中添加对应的列名和友好名称
• 删除列时，从映射表中移除对应条目即可
• 核心逻辑代码无需修改，完全满足后续扩展需求

内容的提问来源于stack exchange，提问作者Antonio