使用 NA 值查找 POSIXct 日期的最小值或最大值

Question

下面的数据包含用于个人 ID（具有重复观察）、 Date和Fate 。

         ID       Date  Fate
1  BHS_1149 2017-04-11   MIA
2  BHS_1154       <NA>  <NA>
3  BHS_1155       <NA>  <NA>
4  BHS_1156       <NA>  <NA>
5  BHS_1157       <NA>  Mort
6  BHS_1159 2017-04-11 Alive
7  BHS_1169 2017-04-11 Alive
8  BHS_1259       <NA>  <NA>
9  BHS_1260       <NA>  <NA>
10 BHS_1262 2017-04-11   MIA
11 BHS_1262 2017-07-05 Alive
12 BHS_1262 2017-12-06 Alive
13 BHS_1262 2017-12-06   MIA
14 BHS_1262 2018-01-17  Mort

对于每个 ID，我想创建一个新列，代表Fate is Alive 时的最小Date或最大Date 。 如果在下面的代码中包含和排除na.rm = T参数，我已经尝试了不同的组合，但仍然收到以下警告。

library(tidyverse)
library(lubridate)

dat %>% 
  group_by(ID) %>%
  mutate(
    #the first or min of Date
    FstSurvey = min(Date),
    LstAlive = max(Date[Fate == "Alive"])) %>%
  as.data.frame()

         ID       Date  Fate  FstSurvey   LstAlive
1  BHS_1149 2017-04-11   MIA 2017-04-11       <NA>
2  BHS_1154       <NA>  <NA>       <NA>       <NA>
3  BHS_1155       <NA>  <NA>       <NA>       <NA>
4  BHS_1156       <NA>  <NA>       <NA>       <NA>
5  BHS_1157       <NA>  Mort       <NA>       <NA>
6  BHS_1159 2017-04-11 Alive 2017-04-11 2017-04-11
7  BHS_1169 2017-04-11 Alive 2017-04-11 2017-04-11
8  BHS_1259       <NA>  <NA>       <NA>       <NA>
9  BHS_1260       <NA>  <NA>       <NA>       <NA>
10 BHS_1262 2017-04-11   MIA 2017-04-11 2017-12-06
11 BHS_1262 2017-07-05 Alive 2017-04-11 2017-12-06
12 BHS_1262 2017-12-06 Alive 2017-04-11 2017-12-06
13 BHS_1262 2017-12-06   MIA 2017-04-11 2017-12-06
14 BHS_1262 2018-01-17  Mort 2017-04-11 2017-12-06

Warning messages:
1: In max.default(numeric(0), na.rm = FALSE) :
  no non-missing arguments to max; returning -Inf
2: In max.default(numeric(0), na.rm = FALSE) :
  no non-missing arguments to max; returning -Inf

代码似乎按预期工作，但我无法解释或避免错误，也无法通过max或min帮助页面找到解决方案。 可重现的代码包含在下面。

dat <- structure(list(ID = c("BHS_1149", "BHS_1154", "BHS_1155", "BHS_1156", 
"BHS_1157", "BHS_1159", "BHS_1169", "BHS_1259", "BHS_1260", "BHS_1262", 
"BHS_1262", "BHS_1262", "BHS_1262", "BHS_1262"), Date = structure(c(1491890400, 
NA, NA, NA, NA, 1491890400, 1491890400, NA, NA, 1491890400, 1499234400, 
1512543600, 1512543600, 1516172400), class = c("POSIXct", "POSIXt"
), tzone = ""), Fate = c("MIA", NA, NA, NA, "Mort", "Alive", 
"Alive", NA, NA, "MIA", "Alive", "Alive", "MIA", "Mort")), row.names = c(NA, 
-14L), .Names = c("ID", "Date", "Fate"), class = "data.frame")

Answer 1

我也喜欢编写不会出错的代码。 这是关于如何在没有警告的情况下进行相同计算的建议。 通过使用有序的first和last而不是min和max，您不会遇到 r interpret max(NULL) 变为 Inf 的奇怪情况。

dat %>% 
  group_by(ID) %>%
  mutate(FstSurvey = first(Date, 
                     order_by = Date),
         LstAlive  = last(Date[Fate == "Alive"], 
                     order_by = Date[Fate == "Alive"]))