使用ifelse条件为日期加月份时返回数值而非日期的问题

需求是根据条件为日期添加月份：当ISO列值为ET时，为日期添加92个月。

单独运行以下代码能得到正确的日期格式结果：

ymd(paste(comb_extract_all$hv007, comb_extract_all$hv006, "01", sep = "-")) %m+% months(92)

返回示例：

[1] "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01"
 [9] "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01"
[17] "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01"
[25] "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01" "2011-04-01"

但将逻辑放入ifelse语句赋值给date列时，返回的是数值而非日期：

comb_extract_all$date <- ifelse(comb_extract_all$ISO == "ET", 
                                ymd(paste(comb_extract_all$hv007, comb_extract_all$hv006, "01", sep="-")) %m+% 
                                  months(92),
                                ymd(paste(comb_extract_all$hv007, comb_extract_all$hv006, "01", sep="-")))

相关数据的dput输出（可见date列异常）：

dput(comb_extract_all[1:5,c(1,5,6,23,24)])
structure(list(hhid = c("        1 27", "        1 27", "        1 27", 
"        1 27", "        1 67"), hv006 = c(8, 8, 8, 8, 8), hv007 = c(2003, 
2003, 2003, 2003, 2003), ISO = c("ET", "ET", "ET", "ET", "ET"
), date = c(15065, 15065, 15065, 15065, 15065)), row.names = c("ETPR61SV.1", 
"ETPR61SV.2", "ETPR61SV.3", "ETPR61SV.4", "ETPR61SV.5"), class = "data.frame")

Base R的ifelse()函数会自动简化返回值的类型，丢失日期对象（Date类）的属性，将其转换为底层的数值表示（从1970-01-01开始的天数）。

提供三种可行的解决方法：

方法1：使用dplyr::if_else() #

dplyr包的if_else()会保留对象的类型属性，比base R的ifelse()更严格：

library(dplyr)
library(lubridate)

comb_extract_all$date <- if_else(comb_extract_all$ISO == "ET",
                                 ymd(paste(comb_extract_all$hv007, comb_extract_all$hv006, "01", sep="-")) %m+% months(92),
                                 ymd(paste(comb_extract_all$hv007, comb_extract_all$hv006, "01", sep="-")))

方法2：先计算基础日期，再通过索引赋值 #

先统一生成基础日期列，再对符合条件的行添加月份：

library(lubridate)

# 先创建基础日期列
comb_extract_all$date <- ymd(paste(comb_extract_all$hv007, comb_extract_all$hv006, "01", sep="-"))
# 对ISO为ET的行添加92个月
comb_extract_all$date[comb_extract_all$ISO == "ET"] <- comb_extract_all$date[comb_extract_all$ISO == "ET"] %m+% months(92)

这种方法避免了ifelse()的类型丢失问题，逻辑更清晰。

方法3：将数值转换回日期类型 #

如果已经用ifelse()生成了数值列，可以手动转换回Date类：

comb_extract_all$date <- as.Date(comb_extract_all$date, origin = "1970-01-01")

内容的提问来源于stack exchange，提问作者Sulz