Make Dealing with Dates a Little Easier!

Introduction

In this computing club mini session, we will cover the lubridate package and learn how to better work with dates and times in R. Lubridate was developed by Garrett Grolemund and Hadley Wickham, and is maintained by Vitalie Spinu. Oftentimes, investigators will provide time/date data in raw form, making it difficult to work with these variables. Conversions to the desirable form are tricky, and time-consuming. The functions in the lubridate package help to streamline and facilitate this process. Lubridate is not part of the tidyverse core (only need it when working with dates/times).

Three possible date/time formats

Date tibbles print this as <date>

Time tibbles print this as <time>

Date-time instant in time, tibbles print this as <dttm> (also called POSIXct in R)

Basics

Parsing

ymd(20190116)

## [1] "2019-01-16"

mdy("1/16/2019")

## [1] "2019-01-16"

today()

## [1] "2019-07-01"

now()

## [1] "2019-07-01 17:27:30 EDT"

Simple functions

exdate <- dmy("16/1/2019")

# Extract month
month(exdate)

## [1] 1

# Extract day of week, set label = TRUE to get actual day in words
wday(exdate, label = T)

## [1] Wed
## Levels: Sun < Mon < Tue < Wed < Thu < Fri < Sat

# Change year of example date, and get new day of the week
year(exdate) <- 2016
wday(exdate, label = T)

## [1] Sat
## Levels: Sun < Mon < Tue < Wed < Thu < Fri < Sat

Applications

Creation

Let’s use the nycflights13 data

#install.packages(nycflights13)
library(nycflights13)

Three scenarios of creating a date/time variable:

String

Must specify correct input and parsing will convert to standard date format

# Covered earlier in the parsing section
ymd("2018-04-08")

## [1] "2018-04-08"

mdy("April 8th, 2018")

## [1] "2018-04-08"

dmy("08-April-2018")

## [1] "2018-04-08"

# Can also work without quotes
ymd(20180408)

## [1] "2018-04-08"

# Create date-time
ymd_hms("2017-01-31 20:11:59")

## [1] "2017-01-31 20:11:59 UTC"

mdy_hm("01/31/2017 08:01")

## [1] "2017-01-31 08:01:00 UTC"

# Force with timezone
ymd(20170131, tz = "UTC")

## [1] "2017-01-31 UTC"

Individual date-time components

flights %>% 
  select(year, month, day, hour, minute)

# Create new variable that combines the five columns using make_datetime
flights %>% 
  select(year, month, day, hour, minute) %>% 
  mutate(departure = make_datetime(year, month, day, hour, minute))

# Combine and create variables for departure, arrival, scheduled departure, scheduled arrival

make_datetime_100 <- function(year, month, day, time) {
  make_datetime(year, month, day, time %/% 100, time %% 100)
}

depart <- flights %>% 
          filter(!is.na(dep_time), !is.na(arr_time)) %>% 
          mutate(
          dep_time = make_datetime_100(year, month, day, dep_time),
          arr_time = make_datetime_100(year, month, day, arr_time),
          sched_dep_time = make_datetime_100(year, month, day, sched_dep_time),
          sched_arr_time = make_datetime_100(year, month, day, sched_arr_time)
          ) %>% 
          select(origin, dest, ends_with("delay"), ends_with("time"))

depart

# Visualize departure times

depart %>% 
  ggplot(aes(dep_time)) + 
  geom_freqpoly(binwidth = 86400)

Existing date/time object

Switch between a date-time and a date --> as_datetime() and as_date()

as_datetime(today())

## [1] "2019-07-01 UTC"

as_date(now())

## [1] "2019-07-01"

Date/times as numeric offsets from Unix Epoch 1970-01-01

#Seconds
as_datetime(60 * 60 * 10)

## [1] "1970-01-01 10:00:00 UTC"

#Days
as_date(365 * 10 + 2)

## [1] "1980-01-01"

Date-time

Graphical exploration

Lets revist the wday() function and the nycflights13 dataset

depart %>% 
  mutate(wday = wday(dep_time, label = TRUE)) %>% 
  ggplot(aes(x = wday)) +
    geom_bar()

And the minute() function

depart %>% 
  mutate(minute = minute(dep_time)) %>% 
  group_by(minute) %>% 
  summarise(
    avg_delay = mean(arr_delay, na.rm = TRUE),
    n = n()) %>% 
  ggplot(aes(minute, avg_delay)) +
    geom_line()

Rounding options

floor_date()

Takes a date-time object and rounds it down to the nearest boundary of the specified time unit

depart %>% 
  count(week = floor_date(dep_time, "week")) %>% 
  ggplot(aes(week, n)) +
    geom_line()

round_date()

Takes a date-time object and rounds it to the nearest value of the specified time unit. Exactly halfway --> round up

depart %>% 
  count(week = round_date(dep_time, "week")) %>% 
  ggplot(aes(week, n)) +
    geom_line()

ceiling_date()

Takes a date-time object and rounds it up to the nearest boundary of the specified time unit

depart %>% 
  count(week = ceiling_date(dep_time, "week")) %>% 
  ggplot(aes(week, n)) +
    geom_line()

update() function

Set multiple values at once

datetime <- ymd_hms("2019-06-20 11:44:39")
update(datetime, year = 2018, month = 8, mday = 10, hour = 5)

## [1] "2018-08-10 05:44:39 UTC"

depart %>% 
  mutate(dep_hour = update(dep_time, yday = 1)) %>% 
  ggplot(aes(dep_hour)) +
    geom_freqpoly(binwidth = 300)

Time spans

durations: measure the exact amount of time between two points

periods: track clock times despite leap years, leap seconds, and day light savings time

intervals: protean summary of the time information between two points

Durations

#Subtraction of two dates --> difftime object (seconds, minutes, hours, days, or weeks)

vicky_age <- today() - ymd(19940819)
vicky_age

## Time difference of 9082 days

Let’s convert vicky_age to a duration using the lubridate package

vicky_age <- as.duration(vicky_age)
vicky_age

## [1] "784684800s (~24.87 years)"

We can use the built in features of associated functions to extract the relevant information that we need:

dseconds(25)

## [1] "25s"

dminutes(50)

## [1] "3000s (~50 minutes)"

dhours(23)

## [1] "82800s (~23 hours)"

ddays(10)

## [1] "864000s (~1.43 weeks)"

dweeks(34)

## [1] "20563200s (~34 weeks)"

dyears(12)

## [1] "378432000s (~11.99 years)"

Durations: give time span in seconds

Adding, subtracting, and multiplying

dyears(50) + dweeks(30) + dhours(10) + dminutes(13)

## [1] "1594980780s (~50.54 years)"

2 * dyears(25)

## [1] "1576800000s (~49.97 years)"

#Tomorrow
today() + ddays(1)

## [1] "2019-07-02"

#Last year
today() - dyears(1)

## [1] "2018-07-01"

Periods

Motivation

ex <- ymd_hms("2017-02-10 14:00:00", tz = "America/New_York")
ex

## [1] "2017-02-10 14:00:00 EST"

ex + ddays(40)

## [1] "2017-03-22 15:00:00 EDT"

Like time spans, but without fixed length in seconds

seconds(25)

## [1] "25S"

minutes(45)

## [1] "45M 0S"

hours(16)

## [1] "16H 0M 0S"

days(40)

## [1] "40d 0H 0M 0S"

months(7)

## [1] "7m 0d 0H 0M 0S"

weeks(23)

## [1] "161d 0H 0M 0S"

years(30)

## [1] "30y 0m 0d 0H 0M 0S"

Adding, subtracting, and multiplying

20 * months(10)

## [1] "200m 0d 0H 0M 0S"

days(30) + hours(4) + minutes(10)

## [1] "30d 4H 10M 0S"

#Leap year
ymd("2016-01-01") + dyears(1)

## [1] "2016-12-31"

#Without taking leap year into account
ymd("2016-01-01") + years(1)

## [1] "2017-01-01"

#DST
ex + ddays(40)

## [1] "2017-03-22 15:00:00 EDT"

#Does not take into account DST
ex + days(40)

## [1] "2017-03-22 14:00:00 EDT"

Application to the nycflights13 dataset

#Extract overnight flights that appear to have arrived before they departed
depart %>% 
  filter(arr_time < dep_time)

depart %>% 
  mutate(
    overnight = arr_time < dep_time,
    arr_time = arr_time + days(overnight * 1),
    sched_arr_time = sched_arr_time + days(overnight * 1)
  )

Intervals

Basic examples

#Interesting result with decimals
years(1) / days(1)

## [1] 365.25

next_year <- today() + years(1)
(today() %--% next_year) / ddays(1)

## [1] 366

#Periods in an interval --> integer division

(today() %--% next_year) %/% days(4)

## [1] 91

Selecting between duration, periods, and intervals

*Simplest = ideal

*Duration –> physical time

*Period –> human times

*Interval –> length of time span in human units

Permitted operations

Further topics

Time zones!

Resouces utilized

https://lubridate.tidyverse.org

https://www.r-statistics.com/2012/03/do-more-with-dates-and-times-in-r-with-lubridate-1-1-0/

https://r4ds.had.co.nz/dates-and-times.html#time-spans

Working with dates and times in R: Lubridate

Victoria Cooley

July 2, 2019