R函数警告排查：'numerical expression has 6 elements: only the first used'

示例数据 #

example_data <- data.frame(
  ID = 1:6,
  Month.Of.birth = c("September", "April", "December", "June", "April", "September"),
  year.of.birth = c(1942, 1942, 1938, 1946, 1944, 1946)
) 

问题描述 #

尝试为每条记录随机分配出生日期的日部分，编写的函数单独运行正常，但传入数据框的整列后出现警告：

Warning in 1:num_days_in_month : numerical expression has 6 elements: only the first used

用户编写的代码如下：

generate_random_day_of_birth <- function(year, month) {
  if (is.na(year) || is.na(month)) {
    return(NA)
  }

  leap_year <- year %% 4 == 0 & (year %% 100 != 0 | year %% 400 == 0)
  
  num_days_in_month <- ifelse(month == "January", 31,
                              ifelse(month == "February" & !leap_year, 28,
                                     ifelse(month == "February" & leap_year, 29,
                                            ifelse(month %in% c("April", "June", "September", "November"), 30, 31))))
  
  random_day <- sample(1:num_days_in_month, 1)
  
  return(random_day)
}

example_data$day.of.birth <- generate_random_day_of_birth(example_data$year.of.birth, example_data$Month.Of.birth)

警告原因 #

1. 函数设计与输入不匹配：你写的函数是针对单个year和month值设计的，但调用时传入了长度为6的向量（整个数据列）。此时num_days_in_month会生成一个长度为6的向量，而R中1:向量这种写法只会取向量的第一个元素来生成序列，忽略其余元素，因此触发警告。
2. 结果不符合预期：不仅会触发警告，sample(1:num_days_in_month, 1)只会基于第一个元素对应的日期范围生成一个随机数，然后将这个数重复赋值给所有6条记录，完全达不到“每条记录随机分配日”的目的。

修正方案 #

方案1：用Vectorize包装函数，使其支持向量输入 #

Vectorize可以将原本处理单个值的函数转换为支持向量输入的函数，自动逐元素处理：

generate_random_day_of_birth <- Vectorize(function(year, month) {
  if (is.na(year) || is.na(month)) {
    return(NA)
  }

  leap_year <- year %% 4 == 0 & (year %% 100 != 0 | year %% 400 == 0)
  
  num_days_in_month <- ifelse(month == "January", 31,
                              ifelse(month == "February" & !leap_year, 28,
                                     ifelse(month == "February" & leap_year, 29,
                                            ifelse(month %in% c("April", "June", "September", "November"), 30, 31))))
  
  random_day <- sample(1:num_days_in_month, 1)
  
  return(random_day)
})

example_data$day.of.birth <- generate_random_day_of_birth(example_data$year.of.birth, example_data$Month.Of.birth)

方案2：用dplyr按行处理 #

如果习惯用tidyverse风格，可以用rowwise()实现逐行计算：

library(dplyr)

example_data <- example_data %>%
  rowwise() %>%
  mutate(day.of.birth = {
    if (is.na(year.of.birth) || is.na(Month.Of.birth)) {
      NA_integer_
    } else {
      leap_year <- year.of.birth %% 4 == 0 & (year.of.birth %% 100 != 0 | year.of.birth %% 400 == 0)
      num_days <- case_when(
        Month.Of.birth == "January" ~ 31,
        Month.Of.birth == "February" & !leap_year ~ 28,
        Month.Of.birth == "February" & leap_year ~ 29,
        Month.Of.birth %in% c("April", "June", "September", "November") ~ 30,
        TRUE ~ 31
      )
      sample(1:num_days, 1)
    }
  }) %>%
  ungroup()

内容的提问来源于stack exchange，提问作者DW1310