Learn

R has three main time types:

• Date — calendar date, no time. as.Date("2026-05-08")
• POSIXct — timestamp with timezone. Stored as seconds since 1970.
• POSIXlt — broken-down time (year, month, day, ...). List-like.

Parsing:

as.Date("2026-05-08")                    # ISO format default
as.Date("05/08/2026", format = "%m/%d/%Y")
as.POSIXct("2026-05-08 14:30:00", tz = "UTC")

Format codes (mostly strftime-compatible):

• %Y 4-digit year, %y 2-digit
• %m month, %d day, %H hour (24h), %I (12h), %p AM/PM
• %M minute, %S second
• %a short weekday, %A full weekday
• %b short month, %B full month

Formatting back to string:

format(Sys.Date(), "%Y-%m-%d")          # ISO
format(Sys.time(), "%A, %B %d at %I:%M %p")

Arithmetic — adding/subtracting durations:

as.Date("2026-05-08") + 7              # one week later
Sys.Date() - as.Date("2020-01-01")     # difftime in days

library(lubridate)
ymd("2026-05-08") + days(30)
ymd("2026-05-08") + months(3)           # lubridate handles month math correctly

Sys.Date() and Sys.time() for now:

Sys.Date()                              # "2026-05-08"
Sys.time()                              # "2026-05-08 14:32:01 UTC"

Components with POSIXlt:

t <- as.POSIXlt(Sys.time())
t$year + 1900                           # year (POSIXlt stores year - 1900!)
t$mon + 1                                # month (0-indexed!)
t$mday                                   # day of month
t$wday                                   # day of week (0 = Sunday)

The 1900 / 0-indexing legacy is a famous footgun.

lubridate (tidyverse) is the modern way:

library(lubridate)

ymd("2026-05-08")                       # parse year-month-day
mdy("05/08/2026")                       # month-day-year
dmy("08-05-2026")                       # day-month-year
ymd_hms("2026-05-08 14:30:00")          # with time

year(today())                            # 2026
month(today())                           # 5
wday(today(), label = TRUE)              # "Fri" (or whatever)

today() %within% interval(ymd("2026-01-01"), ymd("2026-12-31"))   # TRUE

Time series with xts and zoo packages — for financial data, sensors, anything with a regular time index.