R 如何获得具有多个单位的输出时间差

你好

我在as.POSIXct中有两个日期列，格式为YYYY-MM-DD HH:MM:SS。我想得到两者之间的差异，以天小时：秒的格式显示。以下是一些虚拟数据：

    a<-c("2018-03-20 11:52:25 AST", "2018-03-20 12:51:25 AST", "2018-03-20 14:19:04 AST",
"2018-03-21 14:12:12 AST", "2018-03-21 12:09:22 AST", "2018-03-21 15:28:01 AST")

b<-c("2018-04-09 18:39:38 AST", "2018-06-23 19:13:14 AST", "2018-03-20 23:23:03 AST",
     "2018-05-10 21:29:28 AST", "2018-03-22 03:17:23 AST", "2018-05-12 00:19:39 AST")

ab<-data.frame(a,b)

我想得到a和b之间的差值，或者从时间b中减去时间a，得到X天X小时X秒的输出

我在下面使用了difftime，以及不同设置的单位：

ab$time_difference<-difftime(ab$b, ab$a)
ab
                            a                       b   time_difference
     2018-03-20 11:52:25 AST 2018-04-09 18:39:38 AST  486.786944 hours
     2018-03-20 12:51:25 AST 2018-06-23 19:13:14 AST 2286.363611 hours
     2018-03-20 14:19:04 AST 2018-03-20 23:23:03 AST    9.066389 hours
     2018-03-21 14:12:12 AST 2018-05-10 21:29:28 AST 1207.287778 hours
     2018-03-21 12:09:22 AST 2018-03-22 03:17:23 AST   15.133611 hours
     2018-03-21 15:28:01 AST 2018-05-12 00:19:39 AST 1232.860556 hours

---

ab$time\u differencehms库可以在这里提供一些帮助：

library(hms)
as.hms(ab$time_difference, format="%H:%M:S")
# 486:47:13
# 2286:21:49
# 09:03:59
# 1207:17:16
# 15:08:01
# 1232:51:38

有关其他选项，请参见此问题：

以下是上述问题答案中的代码：

Fmt <- function(x) UseMethod("Fmt")
Fmt.difftime <- function(x) {
   units(x) <- "secs"
   x <- unclass(x)
   NextMethod()
}
Fmt.default <- function(x) {
   y <- abs(x)
   sprintf("%s%02d:%02d:%02d:%02d", 
           ifelse(x < 0, "-", ""), # sign
           y %/% 86400,  # days
           y %% 86400 %/% 3600,  # hours 
           y %% 3600 %/% 60,  # minutes
           y %% 60 %/% 1) # seconds
}


a<-c("2018-03-20 11:52:25 AST", "2018-03-20 12:51:25 AST", "2018-03-20 14:19:04 AST",
     "2018-03-21 14:12:12 AST", "2018-03-21 12:09:22 AST", "2018-03-21 15:28:01 AST")

b<-c("2018-04-09 18:39:38 AST", "2018-06-23 19:13:14 AST", "2018-03-20 23:23:03 AST",
     "2018-05-10 21:29:28 AST", "2018-03-22 03:17:23 AST", "2018-05-12 00:19:39 AST")
ab<-data.frame(a,b)

#Passing two dates to  the function(s)
Fmt(as.POSIXct(ab$b)-as.POSIXct(ab$a))
#Passing a time difference in seconds
Fmt(difftime(ab$b, ab$a, units="secs"))

Fmt
require（润滑）
a由于您需要天、小时、分钟、秒，我们可以使用
lubridate
软件包获得此结果：

a<-c("2018-03-20 11:52:25 AST", "2018-03-20 12:51:25 AST", "2018-03-20 14:19:04 AST",
 "2018-03-21 14:12:12 AST", "2018-03-21 12:09:22 AST", "2018-03-21 15:28:01 AST")

b<-c("2018-04-09 18:39:38 AST", "2018-06-23 19:13:14 AST", "2018-03-20 23:23:03 AST",
 "2018-05-10 21:29:28 AST", "2018-03-22 03:17:23 AST", "2018-05-12 00:19:39 AST")

a = as.POSIXct(a)
b = as.POSIXct(b)

library(lubridate)
timespan = interval(ymd_hms(ab[,1]), ymd_hms(ab[,2]))
> as.period(timespan)
[1] "20d 6H 47M 13S"    "3m 3d 6H 21M 49S"  "9H 3M 59S"         "1m 19d 7H 17M 16S"
[5] "15H 8M 1S"         "1m 20d 8H 51M 38S"

使用
sprintf
和模块化算法：

# first, be sure to specify units in difftime, or it will internally
#   choose units for each row
# using 'secs' here since it's the lowest common denominator
# wrapping as.double() to remove the class attribute which will
#   screw up dispatch to Ops below
ab$time_difference <- as.double(difftime(ab$b, ab$a, units = 'secs'))

#  3600 =   60*60 seconds in an hour;
# 86400 = 3600*24 seconds in a day
ab$hms = with(ab, sprintf('%d days; %d hours; %d seconds',
                          time_difference %/% 86400L,
                          (time_difference %% 86400L) %/% 3600L,
                          time_difference %% 3600L))
ab$hms
# [1] "20 days; 6 hours; 2833 seconds" "95 days; 6 hours; 1309 seconds"
# [3] "0 days; 9 hours; 239 seconds"   "50 days; 7 hours; 1036 seconds"
# [5] "0 days; 15 hours; 481 seconds"  "51 days; 8 hours; 3098 seconds"

#首先，确保在difftime中指定单位，否则将在内部指定
#为每行选择单位
#在这里使用“secs”，因为它是最小的公分母
#将包装为.double（）以删除将
#把调度工作搞砸了
ab$time_difference同一天的时差如何等于
-785104d-7H-7m0s
？为此使用
时段对象似乎不是一个理想的想法-“因为时段没有固定的长度”，谢谢。输出中的第二个值为3m和21m。。。小写的m代表毫秒吗？如果是的话，有没有办法去掉毫秒？小写的“m”代表月（因为日期之间的差异很大）。啊，我明白了，谢谢。由于月份的天数不同，有没有办法让月份默认为天？所以我的输出将是天、小时、分钟、秒？@ElizabethSmith:是的，我们可以通过指定格式将月改为天。我已经编辑了我的答案，并在最后添加了一些代码来完成这项工作。谢谢，这是通过将时差列重新格式化为小时：分钟：秒来实现的，但我无法将其格式化为天。我还研究了您链接的问题，并尝试使用该函数，但不太了解如何使用该函数，无法使其重新格式化我的数据列。感谢您的回复和编辑@Dave2e。我已经运行了代码，但不断收到错误消息“error in
*tmp*
$diff:$operator对原子向量无效”。我尝试了is.atomic，每一列都实现了。根据这个问题，我必须使用Fmt（as.POSIXct（ab[“b”]）-as.POSIXct（ab[“a”]），但是当我使用str（ab）时，我会得到错误“不知道如何将'tag_final[“last_pos”]”转换为类“POSIXct”它说这些列在POSIXct中。我这里缺少什么？如果这些列是类POSIXct，那么它们都准备好了一个日期时间对象，您应该使用
difftime
函数。如果您引用列名，则需要使用双括号
Fmt（as.POSIXct（ab[“b”]]）-as.POSIXct（ab[“a”]]）
require(lubridate)

a<-c("2018-03-20 11:52:25 AST", "2018-03-20 12:51:25 AST", "2018-03-20 14:19:04 AST",
     "2018-03-21 14:12:12 AST", "2018-03-21 12:09:22 AST", "2018-03-21 15:28:01 AST")

b<-c("2018-04-09 18:39:38 AST", "2018-06-23 19:13:14 AST", "2018-03-20 23:23:03 AST",
     "2018-05-10 21:29:28 AST", "2018-03-22 03:17:23 AST", "2018-05-12 00:19:39 AST")

# Make df
ab <- data.frame(a = as.POSIXct(a),b = as.POSIXct(b),stringsAsFactors = FALSE)

# Time diff
ab$time_difference <- ab$b - ab$a
ab$time_difference <- as.duration(ab$time_difference)
ab$time_difference 

1 2018-03-20 11:52:25 2018-04-09 18:39:38   1752433s (~2.9 weeks)
2 2018-03-20 12:51:25 2018-06-23 19:13:14 8230909s (~13.61 weeks)
3 2018-03-20 14:19:04 2018-03-20 23:23:03    32639s (~9.07 hours)
4 2018-03-21 14:12:12 2018-05-10 21:29:28  4346236s (~7.19 weeks)
5 2018-03-21 12:09:22 2018-03-22 03:17:23   54481s (~15.13 hours)
6 2018-03-21 15:28:01 2018-05-12 00:19:39  4438298s (~7.34 weeks)

a<-c("2018-03-20 11:52:25 AST", "2018-03-20 12:51:25 AST", "2018-03-20 14:19:04 AST",
 "2018-03-21 14:12:12 AST", "2018-03-21 12:09:22 AST", "2018-03-21 15:28:01 AST")

b<-c("2018-04-09 18:39:38 AST", "2018-06-23 19:13:14 AST", "2018-03-20 23:23:03 AST",
 "2018-05-10 21:29:28 AST", "2018-03-22 03:17:23 AST", "2018-05-12 00:19:39 AST")

a = as.POSIXct(a)
b = as.POSIXct(b)

library(lubridate)
timespan = interval(ymd_hms(ab[,1]), ymd_hms(ab[,2]))
> as.period(timespan)
[1] "20d 6H 47M 13S"    "3m 3d 6H 21M 49S"  "9H 3M 59S"         "1m 19d 7H 17M 16S"
[5] "15H 8M 1S"         "1m 20d 8H 51M 38S"

> as.period(timespan, unit = "day")
[1] "20d 6H 47M 13S" "95d 6H 21M 49S" "9H 3M 59S"      "50d 7H 17M 16S"
[5] "15H 8M 1S"      "51d 8H 51M 38S"

# first, be sure to specify units in difftime, or it will internally
#   choose units for each row
# using 'secs' here since it's the lowest common denominator
# wrapping as.double() to remove the class attribute which will
#   screw up dispatch to Ops below
ab$time_difference <- as.double(difftime(ab$b, ab$a, units = 'secs'))

#  3600 =   60*60 seconds in an hour;
# 86400 = 3600*24 seconds in a day
ab$hms = with(ab, sprintf('%d days; %d hours; %d seconds',
                          time_difference %/% 86400L,
                          (time_difference %% 86400L) %/% 3600L,
                          time_difference %% 3600L))
ab$hms
# [1] "20 days; 6 hours; 2833 seconds" "95 days; 6 hours; 1309 seconds"
# [3] "0 days; 9 hours; 239 seconds"   "50 days; 7 hours; 1036 seconds"
# [5] "0 days; 15 hours; 481 seconds"  "51 days; 8 hours; 3098 seconds"