为何count()、filter()、mutate()在R语言for循环中无法运行？

问题根源 #

问题出在dplyr的默认求值逻辑：它的核心函数（count、filter、mutate等）默认用非标准求值——你直接写count(Q)，它会在数据框里找名叫Q的列，而不是把变量Q里存的字符串（比如Q29、Q30）当作列名来用，自然就会报错说找不到Q列。后面的filter和mutate报错也是同一个逻辑。

一、修正现有代码 #

要让dplyr识别Q里的字符串是列名，需要用准引用（quasiquotation），通过!!sym(Q)把字符串转换成列名引用，告诉dplyr展开这个变量的值而不是当作列名。

1. 修正循环中的代码 #

listDFs <- list()

for (i in 1:2) {
  Q <- names(ch_data[, 1:2][i])
  
  listDFs[[i]] <-  ch_data %>% 
    dplyr::count(!!sym(Q)) %>% 
    mutate(prop = round((prop.table(n) * 100), digits = 2),
           sd = round(sd(prop.table(n)), digits = 2)) %>%
    arrange(!!sym(Q))
}

• sym(Q)：把字符串格式的列名转换成dplyr能识别的符号对象
• !!（双感叹号）：告诉dplyr把符号对象展开成实际的列名，而非当作变量名处理

2. 修正edit部分的代码 #

Q <- names(ch_data[, 1:2][1]) # 示例：取第一列列名

ch_data %>% 
  select(!!sym(Q)) %>% 
  filter(!(!!sym(Q)) %in% 'Não oferta') %>% 
  count(!!sym(Q)) %>% 
  mutate(prop = round((prop.table(n) * 100), digits = 2),
         sd = round(sd(prop.table(n)), digits = 2),
         CH = !!sym(Q),
         QUESTION = rep('mylabel', n())) %>% 
  select(-!!sym(Q)) %>%
  arrange(CH) %>% 
  relocate(QUESTION, CH, .before = everything())

• filter里要给!!sym(Q)加括号，避免优先级问题
• rep('mylabel', n())用n()更稳妥，匹配当前结果的行数

二、更优方案：批量处理替代循环 #

用purrr::map结合函数封装，更符合tidyverse风格，代码更简洁且易维护：

library(purrr)
library(dplyr)

# 定义处理单列的函数
process_single_col <- function(col_name) {
  ch_data %>% 
    select(!!sym(col_name)) %>% 
    filter(!(!!sym(col_name)) %in% 'Não oferta') %>% 
    count(!!sym(col_name)) %>% 
    mutate(prop = round((prop.table(n) * 100), digits = 2),
           sd = round(sd(prop.table(n)), digits = 2),
           CH = !!sym(col_name),
           QUESTION = rep('mylabel', n())) %>% 
    select(-!!sym(col_name)) %>%
    arrange(CH) %>% 
    relocate(QUESTION, CH, .before = everything())
}

# 批量处理指定列
target_cols <- names(ch_data[, 1:2])
listDFs <- map(target_cols, process_single_col)

三、关于你用unlist(Q)解决的原理 #

你写ch_data$QUnlist <- unlist(Q)，本质是把Q对应的列（比如ch_data$Q29）赋值给了新列QUnlist，之后count(QUnlist)自然能找到这个新增列。但这种方法会在原数据框中冗余新列，处理多列时会产生大量无用列，灵活性远不如准引用方案。

测试数据 #

ch_data <- structure(list(Q29 = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 5L, 1L, 1L, 1L, 1L, 4L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 3L, 1L, 1L, 1L, 1L, 1L, 1L, 4L, 1L, 1L, 1L, 1L, 
1L, 1L, 2L, 1L, 5L, 1L, 1L, 1L, 1L, 1L, 1L, 2L), .Label = c("Não oferta", 
"1h - 3h", "Mais de 10h", "50% em LA", "100% em LA"), class = "factor"), 
    Q30 = structure(c(4L, 8L, 3L, 1L, 2L, 2L, 1L, 2L, 1L, 1L, 
    1L, 2L, 5L, 1L, 1L, 1L, 7L, 7L, 1L, 1L, 1L, 1L, 5L, 1L, 1L, 
    2L, 1L, 6L, 2L, 8L, 8L, 2L, 8L, 3L, 7L, 7L, 1L, 2L, 2L, 8L, 
    3L, 2L, 1L, 8L, 3L, 1L, 1L, 1L, 1L, 6L, 3L), .Label = c("Não oferta", 
    "1h - 3h", "4h - 5h", "6h - 8h", "9h - 10h", "Mais de 10h", 
    "50% em LA", "100% em LA"), class = "factor")), row.names = c(NA, 
-51L), class = "data.frame")

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