10.6 Exercises

ds4psy: Exercises (10: Dates and times)

Here are some exercises on creating and computing with dates and times:

Please note

The lubridate package is not part of the core tidyverse packages. Hence, do not forget loading this package if you want to use its commands:

library(tidyverse)      # core tidyverse
library(lubridate)
library(ds4psy, unikn)  # other packages

Note that some key tasks (e.g., computing someone’s age, determining the weekday of some date) occur repeatedly throughout these exercises. If this gets boring, use different solution paths for solving them.

10.6.1 Exercise 1

Reading dates and times

  1. Use the appropriate lubridate function to parse each of the following dates:
d1 <- "January 20, 2020"
d2 <- "2020-Apr-01"
d3 <- "11-Nov-2020"
d4 <- c("July 13 (1969)", "August 23 (1972)", "July 1 (1975)")

                 # Date: 
d5 <- "08/12/10" # Oct 12, 2008
d6 <- d5         # Aug 12, 2010
d7 <- d5         # Oct 08, 2012
  1. Use the appropriate lubridate function to parse each of the following date-times:
t1 <- "2020-11-11 11:11:01"
t2 <- "2020/12/24 07:30"
t3 <- "31:12:20 12:45:59"

t4 <- c("8:05 01/01/2020", "9:20 29/02/2020", "12:30 24/12/2020", "23:58 30/12/2020")

Hint: Note that t4 contains the time component before the date component. To handle this vector, consider creating a tibble and then using dplyr commands for separating its time and date components, and pasting them in reversed order (date before time).

  1. Determine the weekdays of the 7 dates in d4 and t4.

Hint: First combine the seven dates into a vector. Then choose from an abundance of options — the base R function weekdays(), the lubridate function wday(), or the ds4psy function what_wday() — to solve the task.

10.6.2 Exercise 2

Birth dates and times

The table dt_10 (available from ds4psy or rpository.com) contains the birth dates and times of ten non-existent people. Read the data into a tibble dt_10:

# dt_10 <- readr::read_csv("./data/dt_10.csv")  # from local file
# dt_10 <- readr::read_csv("http://rpository.com/ds4psy/data/dt_10.csv")  # online
dt_10 <- ds4psy::dt_10  # from ds4psy

# Show data:
knitr::kable(dt_10, caption = "Data of table `dt_10`.")
Table 10.3: Data of table dt_10.
name day month year hour min sec
Anna 8 8 1994 11 47 57
Beowulf 1 6 1994 5 35 43
Cassandra 14 11 2000 5 58 6
David 17 1 1991 13 3 12
Eva 21 1 2001 21 33 55
Frederic 19 7 2000 13 47 12
Gwendoline 20 9 1996 8 28 37
Hamlet 5 5 1996 17 7 8
Ian 18 8 1996 8 27 17
Joy 18 12 1990 14 44 35
  1. Use base R commands (with “POSIX” specifications) or the corresponding lubridate functions to parse the data of birth dob and time of birth tob as two new columns of dt_10.

Hint: When using base R commands, consider using paste() for creating a character string with appropriate separators from the date- and time-related variables contained in dt_10.

  1. As it turns out, all the people of dt_10 were born in Denmark. Create a second tibble dt_10_2 that considers this fact for the tob variable (e.g., when using the make_datetime() function) and quantify and explain any discrepancies between dt_10$tob and the corresponding variable in dt_10_2.

  2. Use the appropriate lubridate functions to add two columns that specify – given each person’s DOB – the weekday dob_wd (from Monday to Sunday) of their birthday and their current age age_fy in full years (i.e., the numeric value of their age, as an integer).

Hint: Their current age can be computed by subtracting their DOB from today’s date today(). One way of computing their age in full years is by dividing the interval() of their current age by a duration() in the unit of “years.” (Alternatively, rounding can also work.)

10.6.3 Exercise 3

This exercise uses the fame dataset included in the ds4psy package. Actually, the entries of the dataset were populated by the submissions of previous students. So think carefully about your entries — they might end up in the dataset studied by future generations of students.

Add to fame

  1. Pick at least four famous people — some of which are still alive, some of which have already died — and enter their name, area of occupation, date of birth (DOB), and date of death (DOD, if deceased) in a tibble fame, in analogy to the following:
fame <- tibble(name = c("Napoleon Bonaparte", "Jimi Hendrix", "Michael Jackson", "Frida Kahlo", 
                        "Angela Merkel", "Kobe Bryant", "Lionel Messi", "Zinedine Zidane"),
               area = c("politics", "guitarist/music", "singer/music", "arts/painter", 
                        "politics", "basketball/sports", "football/sports", "football/sports"), 
               DOB = c("August 15, 1769", "November 27, 1942", "August 29, 1958", "July 06, 1907", 
                       "July 17, 1954", "August 23, 1978", "June 24, 1987", "June 23, 1972"),
               DOD = c("May 05, 1821", "September 18, 1970", "June 25, 2009", "July 13, 1954", 
                       NA, "January 26, 2020", NA, NA))

knitr::kable(fame, caption = "Basic info on some famous people.")
Table 10.4: Basic info on some famous people.
name area DOB DOD
Napoleon Bonaparte politics August 15, 1769 May 05, 1821
Jimi Hendrix guitarist/music November 27, 1942 September 18, 1970
Michael Jackson singer/music August 29, 1958 June 25, 2009
Frida Kahlo arts/painter July 06, 1907 July 13, 1954
Angela Merkel politics July 17, 1954 NA
Kobe Bryant basketball/sports August 23, 1978 January 26, 2020
Lionel Messi football/sports June 24, 1987 NA
Zinedine Zidane football/sports June 23, 1972 NA

Note: Please remember to enter any rare and unusual symbols as Unicode characters (see Section 9.2.2).

  1. Use the appropriate lubridate functions to replace the DOB and DOD variables in fame by corresponding dob and dod variables of type “Date.”

  2. Add two variables to fame that specify the weekday (from “Monday” to “Sunday”) of their birth (dob_wd) and — if applicable — of their death (dob_wd).

  3. Add a variable age_days that computes their age in days (relative to today’s date). Then compute two more variables age_yr1 and age_yr2 that determines their age in years (as a decimal number) in two different ways. Finally, add a variable age_fy that specifies their current age (in full years) as an integer (i.e., what they would say if they truthfully responded to the question “How old are you today?”).

  4. Correct your previous age_fyr variable so that — for those people who have already died — it should remain at the age at which they died (i.e., dead people do not age further).

Note: The Pantheon site contains more sophisticated records of famous individuals (Yu, Ronen, Hu, Lu, & Hidalgo, 2016). Large datasets can be downloaded at https://pantheon.world/data/datasets.

10.6.4 Exercise 4

Time conversions

  1. Define a time point of the New Year fireworks in Sydney, Australia, as “2021-01-01 00:00:01” (including time zone information).

  2. Predict and explain the results of the following commands in your own words.

with_tz(t_fw,  tz = "Europe/Berlin")
#> [1] "2020-12-31 14:00:01 CET"
force_tz(t_fw, tz = "Europe/Berlin")
#> [1] "2021-01-01 00:00:01 CET"
  1. Predict and explain the outcome of the following commands.
t_fw -  with_tz(t_fw, tz = "Europe/Berlin")
t_fw - force_tz(t_fw, tz = "Europe/Berlin")

Hint: This is possible without actually running them (after having done 2.).

10.6.5 Exercise 5

Hoop times

This exercise uses the lakers dataset included in lubridate (originally from http://www.basketballgeek.com/data/), which contains play-by-play statistics of each Los Angeles Lakers (LAL) basketball game in the 2008/2009 season of the NBA. (See ?lakers for details.)

  1. Select only those games against the Dallas Mavericks (abbreviated as “DAL”) and save the corresponding data as a tibble LAL_DAL.

  2. Use your tidyverse knowledge acquired so far to answer some basic questions about those games:

    • How many such (home vs. away) games exist?
    • On which dates were they played?
    • What were their scores? Who won the game?

Hint: All these questions can be answered with a single dplyr pipe.

  1. Create and add the following date and time variables to LAL_DAL:

    • date should be a variable of type “Date” (rather than a character string)
    • t_clock should represent the time shown on the clock (as a period)
    • t_psec should represent the time elapsed in the current period (a duration in seconds)
    • t_game should represent the time elapsed in the game overall (as a duration).

Hint: An NBA game consists of 4 periods, each of which lasts 12 minutes (i.e., each game’s time should add up to a total of 48 minutes).

  1. Prominent players:

    • For which individual player on each team do the data record the highest number of events?
    • How many points did each of these two players score (over all games)?
    • What would it take to compute the time difference between all recorded events for these two players as lubridate intervals?
    • Bonus task: Compute these intervals for each of these two players.
    • What would it take to compute the time difference between all recorded events for these two players as lubridate intervals?

  1. Cumulative points per game:

    • Compute and add a variable for the cumulative point_total of each game and team.
    • Compute the final score f_score of each game and team (and compare your result to the one obtained to answer 2. above).
    • Plot the (cumulative) point_total for each game per team as a function of t_game.

Please note: This dataset and questions like the ones asked here are a good illustration of a possible Data science project. At this point, you should be starting to think about datasets and questions for your own project. (See Appendix C for some guidelines for and the scope of a successful data science projects.)

10.6.6 Exercise 6

DOB and study times

The dataset exp_num_dt (available in the ds4psy package or as a CSV-file from rpository.com) contains the birth dates and study participation times of 1000 ficticious, but surprisingly friendly people.

We read the data file into a tibble dt and select only its date-related variables:

# dt <- readr::read_csv("http://rpository.com/ds4psy/data/dt.csv")  # online 
dt <- ds4psy::exp_num_dt  # ds4psy package
# dt

# Select only its date-time related variables:
dt_t <- dt %>% select(name:byear, t_1, t_2)

# Check:
# dt  # 1000 x 7
knitr::kable(head(dt_t), caption = "Time-related variables of table `dt`.")
Table 10.5: Time-related variables of table dt.
name gender bday bmonth byear t_1 t_2
I.G. male 14 12 1968 2020-01-16 11:00:58 2020-01-16 11:32:21
O.B. male 10 4 1974 2020-01-17 14:11:07 2020-01-17 15:05:14
M.M. male 28 9 1987 2020-01-16 10:06:06 2020-01-16 10:51:47
V.J. female 15 2 1978 2020-01-10 10:06:04 2020-01-10 10:39:48
O.E. male 18 5 1985 2020-01-20 09:23:51 2020-01-20 10:11:36
Q.W. male 1 3 1968 2020-01-13 11:10:09 2020-01-13 11:54:07

  1. The variables bday, bmonth, and byear contain each participant’s date of birth.

    • Compute a variable DOB that summarizes bday, bmonth, and byear (in a “Date” variable) and a variable bweekday that shows the weekday of each participant’s DOB (as a chacter variable).

Hint: A base R solution is about as long as the dplyr/lubridate solution.

  1. What would each participant respond to the question

    • “How old are you?”

(i.e., what was each person’s age in completed years, when starting the study in January 2020)? Verify your result for those participants who took part in the study on their birthday.

Hint: This task requires considering both DOB and t_1 (to check whether the person already celebrated his or her birthday in the current year when starting the study at the time t_1).

  1. The time variables t_1 and t_2 indicate the start and end times of each person’s participation in this study.

    • Compute the duration of each person’s participation (in minutes and seconds) and plot the distribution of the resulting durations (e.g., as a histogram).

  1. The study officially only ran for five days — from “2020-01-13” to “2020-01-18” — and should only include participants that responded in up to one hour (60 minutes).

    • Add a filter variable valid that enforces these criteria (i.e., allows filtering out participants with other dates and durations longer than 60 minutes).

  1. Finally, we can compute some basic descriptives of the participants considered to be valid:

    • How many participants remain in the sample of valid data?
    • What is their average height and g_iq score?

10.6.7 Exercise 7

Bonus task: Evaluating time differences

This exercise creates random time differences and compares the results of computing them in two different ways.

  1. Use the sample_time() function of ds4psy to generate vectors of N random starting times and N random end times.

  2. Compute and compare the time difference between both vectors for various units of time. Specifically, compare the solutions of the diff_times() function of ds4psy with the corresponding lubridate solution (using time intervals and periods).

  3. Continue comparing the results of both solution methods until you find some examples with different solutions for the same time difference. Can you explain the discrepancies?

Hint: Here is a possible setup for an investigation of this type:

# Parameters:
N <- 10
t1 <- "2020-01-01 00:00:00"
t2 <- Sys.time()

# Random time vectors:
t_start <- ds4psy::sample_time(from = t1, to = t2, size = N)
t_end   <- ds4psy::sample_time(from = t1, to = t2, size = N)

# in months:
ds4psy::diff_times(t_start, t_end, unit = "months", as_character = FALSE)
lubridate::as.period(lubridate::interval(t_start, t_end), unit = "months")

# in days:
ds4psy::diff_times(t_start, t_end, unit = "days", as_character = FALSE)
lubridate::as.period(lubridate::interval(t_start, t_end), unit = "days")

This concludes our exercises on creating and computing with dates and times.

References

Yu, A. Z., Ronen, S., Hu, K., Lu, T., & Hidalgo, C. A. (2016). Pantheon 1.0, a manually verified dataset of globally famous biographies. Scientific Data, 3(1), 1–16. https://doi.org/10.1038/sdata.2015.75