分组日期和时间

【问题标题】：Grouping dates and times分组日期和时间
【发布时间】：2020-01-31 13:13:35
【问题描述】：

是否有 R 包为所有典型时间单位（秒、分钟等）提供日期和时间分组？

或者换一种说法：如何将特定时间单位截断为倍数？

背景

我经常需要将 HTTP 请求的响应分组为 15 秒的时间间隔。

函数lubridate::floor_date() 和lubridate::ceiling_date() 让我开始了，但是

我想出了一些基于模 (%%) 的东西，但感觉就像我正在重新发明关于操作原子时间组件的轮子。

函数定义

library(magrittr)

group_date <- function(
  x,
  interval = 15,
  unit = c(
    "seconds",
    "minutes",
    "hours",
    "days",
    "weeks",
    "months",
    "bimonths",
    "quarters",
    "seasons",
    "halfyears",
    "years"
  )
) {
  # Validate units:
  unit <- match.arg(unit)

  # Possibly base units on "basic units" as {lubridate} does:
  # parsed_unit <- lubridate:::parse_period_unit(unit)
  # n <- parsed_unit$n
  # basic_unit <- lubridate:::standardise_period_names(parsed_unit$unit)

  if (unit %in% c("bimonths", "halfyears", "season")) {
    stop(stringr::str_glue("Unit '{unit}' not supported yet"))
  }
  # No clue how these would need to be handled yet

  # Extract unit value:
  unit_value <- dplyr::case_when(
    unit == "seconds" ~ as.numeric(lubridate::second(x)),
    unit == "minutes" ~ as.numeric(lubridate::minute(x)),
    unit == "hours" ~ as.numeric(lubridate::hour(x)),
    unit == "days" ~ as.numeric(lubridate::day(x)),
    unit == "weeks" ~ as.numeric(lubridate::isoweek(x)),
    unit == "months" ~ as.numeric(lubridate::month(x)),
    unit == "quarters" ~ as.numeric(lubridate::quarter(x)),
    unit == "year" ~ as.numeric(lubridate::year(x))
  )

  offset_factor <- dplyr::case_when(
    unit == "seconds" ~ 1,
    unit == "minutes" ~ 60,
    unit == "hours" ~ 60 * 60,
    unit == "days" ~ 60 * 60 * 24,
    unit == "weeks" ~ NA_real_, # Seconds per week -> no clue how to do that,
    unit == "months" ~ NA_real_, # Seconds per month -> no clue how to do that
    unit == "quarters" ~ NA_real_, # Seconds per quarter -> no clue how to do that
    unit == "year" ~ NA_real_ # Seconds per year -> no clue how to do that
  )

  # Calculate time offset to lower group boundary:
  time_offset <- unit_value %% interval

  # Apply offset:
  x - (time_offset * offset_factor)
}

分组秒数

x <- c(
  "2020-01-31 13:01:14",
  "2020-01-31 13:01:15",
  "2020-01-31 13:01:16",
  "2020-01-31 13:01:29",
  "2020-01-31 13:01:30",
  "2020-01-31 13:01:31",
  "2020-01-31 13:01:44",
  "2020-01-31 13:01:45",
  "2020-01-31 13:01:46",
  "2020-01-31 13:01:59",
  "2020-01-31 13:02:00",
  "2020-01-31 13:02:01"
) %>%
  lubridate::ymd_hms()

x %>% group_date()
#>  [1] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:15 UTC"
#>  [3] "2020-01-31 13:01:15 UTC" "2020-01-31 13:01:15 UTC"
#>  [5] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [7] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:45 UTC"
#>  [9] "2020-01-31 13:01:45 UTC" "2020-01-31 13:01:45 UTC"
#> [11] "2020-01-31 13:02:00 UTC" "2020-01-31 13:02:00 UTC"

x %>% group_date(30)
#>  [1] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [3] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [5] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [7] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [9] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#> [11] "2020-01-31 13:02:00 UTC" "2020-01-31 13:02:00 UTC"

x %>% group_date(45)
#>  [1] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [3] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [5] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [7] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:45 UTC"
#>  [9] "2020-01-31 13:01:45 UTC" "2020-01-31 13:01:45 UTC"
#> [11] "2020-01-31 13:02:00 UTC" "2020-01-31 13:02:00 UTC"

x %>% group_date(60)
#>  [1] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [3] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [5] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [7] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [9] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#> [11] "2020-01-31 13:02:00 UTC" "2020-01-31 13:02:00 UTC"

分组分钟数

x <- c(
  "2020-01-31 13:04:00",
  "2020-01-31 13:05:00",
  "2020-01-31 13:06:00",
  "2020-01-31 13:29:00",
  "2020-01-31 13:30:00",
  "2020-01-31 13:31:00",
  "2020-01-31 13:44:00",
  "2020-01-31 13:45:00",
  "2020-01-31 13:46:00"
) %>%
  lubridate::ymd_hms()

x %>% group_date(15, "minutes")
#> [1] "2020-01-31 13:00:00 UTC" "2020-01-31 13:00:00 UTC"
#> [3] "2020-01-31 13:00:00 UTC" "2020-01-31 13:15:00 UTC"
#> [5] "2020-01-31 13:30:00 UTC" "2020-01-31 13:30:00 UTC"
#> [7] "2020-01-31 13:30:00 UTC" "2020-01-31 13:45:00 UTC"
#> [9] "2020-01-31 13:45:00 UTC"

x %>% group_date(30, "minutes")
#> [1] "2020-01-31 13:00:00 UTC" "2020-01-31 13:00:00 UTC"
#> [3] "2020-01-31 13:00:00 UTC" "2020-01-31 13:00:00 UTC"
#> [5] "2020-01-31 13:30:00 UTC" "2020-01-31 13:30:00 UTC"
#> [7] "2020-01-31 13:30:00 UTC" "2020-01-31 13:30:00 UTC"
#> [9] "2020-01-31 13:30:00 UTC"

x %>% group_date(45, "minutes")
#> [1] "2020-01-31 13:00:00 UTC" "2020-01-31 13:00:00 UTC"
#> [3] "2020-01-31 13:00:00 UTC" "2020-01-31 13:00:00 UTC"
#> [5] "2020-01-31 13:00:00 UTC" "2020-01-31 13:00:00 UTC"
#> [7] "2020-01-31 13:00:00 UTC" "2020-01-31 13:45:00 UTC"
#> [9] "2020-01-31 13:45:00 UTC"

编辑

刚刚遇到hms::trunc_hms()。

似乎给了我secs = 15 和secs = 30 所需的东西，所以它可以解决我眼前的问题。但我看不出它对于秒以外的时间单位是如何工作的：

library(magrittr)

x <- c(
  "2020-01-31 13:01:14",
  "2020-01-31 13:01:15",
  "2020-01-31 13:01:16",
  "2020-01-31 13:01:29",
  "2020-01-31 13:01:30",
  "2020-01-31 13:01:31",
  "2020-01-31 13:01:44",
  "2020-01-31 13:01:45",
  "2020-01-31 13:01:46",
  "2020-01-31 13:01:59",
  "2020-01-31 13:02:00",
  "2020-01-31 13:02:01"
) %>%
  lubridate::ymd_hms()

x %>% hms::trunc_hms(15)
#>  [1] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:15 UTC"
#>  [3] "2020-01-31 13:01:15 UTC" "2020-01-31 13:01:15 UTC"
#>  [5] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [7] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:45 UTC"
#>  [9] "2020-01-31 13:01:45 UTC" "2020-01-31 13:01:45 UTC"
#> [11] "2020-01-31 13:02:00 UTC" "2020-01-31 13:02:00 UTC"

x %>% hms::trunc_hms(30)
#>  [1] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [3] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [5] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [7] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [9] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#> [11] "2020-01-31 13:02:00 UTC" "2020-01-31 13:02:00 UTC"

x %>% hms::trunc_hms(45)
#>  [1] "2020-01-31 13:00:45 UTC" "2020-01-31 13:00:45 UTC"
#>  [3] "2020-01-31 13:00:45 UTC" "2020-01-31 13:00:45 UTC"
#>  [5] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [7] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#>  [9] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"
#> [11] "2020-01-31 13:01:30 UTC" "2020-01-31 13:01:30 UTC"

x %>% hms::trunc_hms(60)
#>  [1] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [3] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [5] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [7] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#>  [9] "2020-01-31 13:01:00 UTC" "2020-01-31 13:01:00 UTC"
#> [11] "2020-01-31 13:02:00 UTC" "2020-01-31 13:02:00 UTC"

编辑 2

查看使用cut()的建议

x %>%
  tibble::enframe() %>%
  dplyr::mutate(
    grp = value %>% group_date(15)
  ) %>% 
  dplyr::group_by(
    grp_2 = cut(as.POSIXct(value, '%Y-%m-%d %H:%M:%S'), '15 secs')
  )
# # A tibble: 12 x 4
# # Groups:   grp [4]
# name value                   grp_2                   grp                
# <int> <dttm>                  <dttm>                  <fct>              
#   1     1 2020-01-31 13:01:14.000 2020-01-31 13:01:00.000 2020-01-31 13:01:14
# 2     2 2020-01-31 13:01:15.000 2020-01-31 13:01:15.000 2020-01-31 13:01:14
# 3     3 2020-01-31 13:01:16.000 2020-01-31 13:01:15.000 2020-01-31 13:01:14
# 4     4 2020-01-31 13:01:29.000 2020-01-31 13:01:15.000 2020-01-31 13:01:29
# 5     5 2020-01-31 13:01:30.000 2020-01-31 13:01:30.000 2020-01-31 13:01:29
# 6     6 2020-01-31 13:01:31.000 2020-01-31 13:01:30.000 2020-01-31 13:01:29
# 7     7 2020-01-31 13:01:44.000 2020-01-31 13:01:30.000 2020-01-31 13:01:44
# 8     8 2020-01-31 13:01:45.000 2020-01-31 13:01:45.000 2020-01-31 13:01:44
# 9     9 2020-01-31 13:01:46.000 2020-01-31 13:01:45.000 2020-01-31 13:01:44
# 10    10 2020-01-31 13:01:59.000 2020-01-31 13:01:45.000 2020-01-31 13:01:59
# 11    11 2020-01-31 13:02:00.000 2020-01-31 13:02:00.000 2020-01-31 13:01:59
# 12    12 2020-01-31 13:02:01.000 2020-01-31 13:02:00.000 2020-01-31 13:01:59

由reprex package (v0.3.0) 于 2020 年 1 月 31 日创建

【问题讨论】：

  • 你可以使用cut()。你见过this吗？
  • @Sotos 似乎对我不起作用。当我运行 x %&gt;% cut(breaks = "15 min") 时，我得到一个带有 Levels: 2020-01-31 13:01:00 的因子向量
  • 不是那样...试试enframe(x) %&gt;% group_by(grp = cut(as.POSIXct(value, '%Y-%m-%d %H:%M:%S'), '15 secs'))
  • @Sotos 好吧......很抱歉在被指向cut() 后没有建立明显的联系。不过，仍然没有给我想要的东西（见编辑#2）
  • 哦，您希望它始终在 H:00:00 开始吗​​？

标签： r datetime grouping lubridate iso8601


【解决方案1】：

这个简单的函数如何舍入到任何给定的秒数、分钟数、小时数、天数或周数？如果您想在特定日期和时间开始您的块，则有一个可选的默认起始时间。 “单位”的参数是匹配的，因此您可以缩写。默认为秒

time_group <- function(times, intervals, since = as.POSIXct("2000-01-01"), 
                       units = c("secs", "mins", "hours", "days", "weeks"))
{
  all_units <- c("secs", "mins", "hours", "days", "weeks")
  units     <- match.arg(units, all_units)
  intervals <- intervals * c(1, 60, 3600, 86400, 604800)[match(units, all_units)]
  cuts      <- intervals * floor(as.numeric(difftime(times, since, units = "secs"))/intervals)

  return(as.POSIXct(cuts, origin = since))
}

这允许您这样做：

# Units default to seconds so this groups by 15 seconds at a time
time_group(x, 15)
#>  [1] "2020-01-31 13:01:00 GMT" "2020-01-31 13:01:15 GMT" "2020-01-31 13:01:15 GMT"
#>  [4] "2020-01-31 13:01:15 GMT" "2020-01-31 13:01:30 GMT" "2020-01-31 13:01:30 GMT"
#>  [7] "2020-01-31 13:01:30 GMT" "2020-01-31 13:01:45 GMT" "2020-01-31 13:01:45 GMT"
#> [10] "2020-01-31 13:01:45 GMT" "2020-01-31 13:02:00 GMT" "2020-01-31 13:02:00 GMT"

# We have used argument matching so we can abbreviate minutes to "m"
time_group(x, 1, units = "m")
#>  [1] "2020-01-31 13:01:00 GMT" "2020-01-31 13:01:00 GMT" "2020-01-31 13:01:00 GMT"
#>  [4] "2020-01-31 13:01:00 GMT" "2020-01-31 13:01:00 GMT" "2020-01-31 13:01:00 GMT"
#>  [7] "2020-01-31 13:01:00 GMT" "2020-01-31 13:01:00 GMT" "2020-01-31 13:01:00 GMT"
#> [10] "2020-01-31 13:01:00 GMT" "2020-01-31 13:02:00 GMT" "2020-01-31 13:02:00 GMT"

【讨论】：

  • 是的，我认为这与hms::trunc_hms() 使用的基本方法相同：trunc(as.numeric(x)/secs) * secs
  • 但是，当我想分组/截断为该时间单位的倍数时，我如何将其推广到其他时间单位？
猜你喜欢
  • 2020-02-21
  • 1970-01-01
  • 1970-01-01
  • 1970-01-01
  • 2021-11-08
  • 1970-01-01
  • 1970-01-01
  • 1970-01-01
  • 2020-09-09
相关资源
最近更新 更多
热门标签
Java Python linux javascript Mysql C# Docker 算法 前端 SpringBoot Redis Vue spring 设计模式 .net core .net kubernetes c++ 数据库 数据结构 大数据 js 机器学习 微服务 Android Go 程序员 面试 JVM ASP.net core 云原生 人工智能 后端 PHP git CSS golang k8s Nginx Django mybatis 深度学习 多线程 React 架构 devops 爬虫 云计算 Spring Boot LeetCode