16 Introduction to Working with Dates in R

16.1 Introduction

In many data files, the date or time of day will be an important variable. In this introductory tutorial, we will learn some basics on to handles dates.

A Reminder: Why do <date> objects even matter? Compare the following two plots: one made where the date is in <chr> form and the other where date is in its appropriate <date> form.

library(tidyverse)
library(lubridate)

animal_crossing <- read_csv("data/animal_crossing_holidays.csv") 

animal_crossing %>%
ggplot(data = ., aes(x = Date1, y = Holiday)) +
  geom_point()

animal_crossing %>% 
  mutate(Date_test_plot = dmy(Date1)) %>%
  ggplot(data = ., aes(x = Date_test_plot, y = Holiday)) +
  geom_point()

In which plot does the ordering on the x-axis make more sense?

16.2 Dates with lubridate

Goals:

  • use lubridate functions to convert a character variable to a <date> variable.
  • use lubridate functions to extract useful information from a <date> variable, including the year, month, day of the week, and day of the year.

16.2.1 Converting Variables to <date>

The lubridate package is built to easily work with Date objects and DateTime objects.

To begin, here are a few basic functions today(), which prints today’s date, and now() prints today’s date and time.

today()
now()

There are a number of built-in functions to convert character strings to Dates and Times.

16.2.1.1 Parsing Dates and Times

  • ymd(): Parses dates in the format of “year-month-day” and returns a datetime object.
  • dmy(): Parses dates in the format of “day-month-year” and returns a datetime object.
  • mdy(): Parses dates in the format of “month-day-year” and returns a datetime object.
  • hm(): Parses times in the format of “hour-minute” and returns a time object.
  • hms(): Parses times in the format of “hour-minute-second” and returns a datetime object.

Here is a quick example showing what they do.

# dates in different formats

d1 <- "2023-04-19"
d2 <- "19-04-2023"
d3 <- "04-19-2023"

ymd(d1)
dmy(d2)
mdy(d3)

mdy(d2) # fails to parse b/c no month 19
# Parse time in different formats
time_hm <- hm("10:15")
time_hms <- hms("10:15:30")

As seen before, these also work on variables within data frames (tibbles).

animal_crossing %>%
  mutate(Date1_v2 = dmy(Date1)) %>%
  relocate(Date1_v2)

16.2.1.2 year(), month(), and mday()

The functions year(), month(), and mday() can grab the year, month, and day of the month, respectively, from a <date> variable. Like the forcats functions, these will almost always be paired with a mutate() statement because they will create a new variable.

Notice in the animal crossings data there are a number of variables related to these aspects. Here is how they were created.

# starting fresh
animal_crossing2 <- animal_crossing %>% select(Holiday, Date1)

# recreate initial 
animal_crossing2 %>%
  mutate(
    Date = dmy(Date1),
    Month = month(Date),
    Year = year(Date),
    Day = mday(Date),
    Month2 = month(Date, label = TRUE, abbr = FALSE),
    
    # a few extras
    Day_in_year = yday(Date),
    Day_of_week = wday(Date, label = TRUE, abbr = TRUE),
    week_of_year = week(Date)
  )

16.3 Using parse_date from the readr package

Another common way to work with dates is to use the parse_date (and parse_date_time function from lubridate). This usually requires us to identify a format for the date (or date-time) structure.

For example, if you have a date in the format “2023-04-18 09:19:59”, you can use the following format string to parse it using the parse_date() function:

date_string <- "2023-04-18 09:19:59"
date <- parse_date(date_string, 
                   format = "%Y-%m-%d %H:%M:%S"
                   )
date

Below is a table of common formats.

Format Description
%d Day of the month as a number (01-31).
%m Month as a number (01-12).
%Y Year with century (as a four digit number).
%y Year without century (00-99).
%H Hour (24-hour clock) as a decimal number (00-23).
%I Hour (12-hour clock) as a decimal number (01-12).
%M Minute as a decimal number (00-59).
%S Second as a decimal number (00-59).
%z Time zone offset from UTC (e.g., “-0800”).
%Z Time zone name.

16.4 Another Example

We have data on flights originating from New York airports in November 2022.

url <- "https://raw.githubusercontent.com/iramler/stat234/main/notes/data/ny_airports_nov2022.csv"
ny_airports <- read_csv(url)
ny_airports %>% mutate(FL_DATE = str_remove(FL_DATE, " 12:00:00 AM")) -> ny_airports

First, lets reduce this data into just the four airports in Albany, Buffalo, Rochester, and Syracuse.

airports_to_use = paste( c('Albany', 'Buffalo', 'Rochester', 'Syracuse'), "NY", sep = ", ")
airports_to_use

upstate_airports <- 
  ny_airports %>%
    filter(ORIGIN_CITY_NAME %in% airports_to_use)

Now convert the FL_DATE variable from a <chr> to a <date>.

upstate_airports <-
  upstate_airports %>%
  mutate(Flight_Date = mdy(FL_DATE))

  • Calculate the average delay time for each airport.

  • Calculate the proportion of flights delayed for each airport.

upstate_airports %>%
  group_by(ORIGIN_CITY_NAME) %>%
  summarise(
    avgDelay = mean(DEP_DELAY, na.rm=TRUE),
    propDelay = mean( (DEP_DELAY > 0) , na.rm= TRUE )
  )
  • Which day of the week has the most flights?
upstate_airports %>%
  mutate(day.of.week = wday(Flight_Date,
                            label = TRUE,
                            abbr = TRUE
                            ) ) %>%
  group_by(day.of.week) %>%
  summarise(
    n_flights = n()
  ) %>%
  slice_max(n_flights, n = 1)

Try out this plot.

library(ggTimeSeries)
upstate_airports %>%
  group_by(Flight_Date, ORIGIN_CITY_NAME) %>%
  summarise(avg_daily_delay = mean(DEP_DELAY, na.rm=TRUE)) %>%
  ungroup() %>%
 ggplot_calendar_heatmap("Flight_Date",
                        "avg_daily_delay", 
                        ) + 
  facet_wrap(~ORIGIN_CITY_NAME) +
  theme(legend.position = "top") +
  scale_fill_continuous(low = 'green', high = 'red') +
  labs(y = "Day of Week", y = "Month", fill = "Average Delay (min)") +
  coord_flip()

16.5 Another Fun Example

https://trends.google.com/trends/explore?geo=US&hl=en

library(gtrendsR)

search_terms <- c("pumpkin spice","cold brew")

mysearch <- gtrends(
  keyword = search_terms, onlyInterest = TRUE,
  #time = "all", # this will let us get the data from 2004 - present 
  time = "2013-04-18 2023-04-18"
)

coffee_df <- mysearch[[1]]

head(coffee_df)
tail(coffee_df)
  • Make a plot of your Popularity variables through time.