Lubridate: Add a ym() function for year-month "dates" in spirit of yq()?

Created on 29 Jan 2018 · 27Comments · Source: tidyverse/lubridate

As lubridate already has a handy yq() function for parsing year-quarter "dates", would you consider adding ym(), the equivalent for numeric year-month "dates"?

Date such as this crop up all the time in applied sciences, where the exact date of sampling is not recorded or neccessary or the data have been aggregated to a monthly time step, eg monthly climate anomolies, and provided in that format (i.e. without any day component).

parse_date_time knows how to handle such inputs

> parse_date_time("10/2001", "mY")
[1] "2001-10-01 UTC"

I'm thinking of something along the lines of (I'm new to lubridate so I'm not sure I follow the meaning of the orders argument just yet):

ym <- function (..., quiet = FALSE, tz = NULL, locale = Sys.getlocale("LC_TIME")) {
  lubridate:::.parse_xxx(..., orders = "ym", quiet = quiet, tz = tz, locale = locale, 
                    truncated = 0)
}

my <- function (..., quiet = FALSE, tz = NULL, locale = Sys.getlocale("LC_TIME")) {
  lubridate:::.parse_xxx(..., orders = "my", quiet = quiet, tz = tz, locale = locale, 
                    truncated = 0)
}

with expected output

> my("10-2001")
[1] "2001-10-01"
> ym("2010-10")
[1] "2010-10-01"

Optionally, accepting an argument fraction (or equivalent) would allow user to control the day part that is added: would return date + days(fraction * days_in_month)?

Source

gavinsimpson

👍7

Most helpful comment

I will add it soonish.

vspinu on 13 Nov 2019

👍17 ❤6

All 27 comments

I don't think those functions are that useful, especially that the semantics of my and ym are likely not to be clear for a casual reader. Plus one would need yd, md etc. for consistency.

You can already do such partial parsing in a number of ways:

> ymd("2001-01", truncated = 1)
[1] "2001-01-01"
> myd("01-2001", truncated = 1)
[1] "2001-01-01"
> ymd(paste0("2001-01", "-03"))
[1] "2001-01-03"
> parse_date_time("2001-01", "ym")
[1] "2001-01-01 UTC"
> parse_date_time2("2001-01", "Ym")
[1] "2001-01-01 UTC"

For day part one can do x + days(3) or ceiling_date(x, "3d") etc.

vspinu on 29 Jan 2018

👍8 🚀1

I could use this function as well.

zenteknet on 29 Dec 2018

I was looking for this function today. Seems like all possible combinations of dates should be covered to make the library universally useful.

johnvbusch on 15 Feb 2019

👍8

+1 in favor of this function. Writing out ugly solutions like mutate(month = paste0(year(date_col), "-", month(date_col))) is inelegant and seems contrary to the spirit of this package.

jenitivecase on 1 Mar 2019

👍18

Just tossing my vote for support on this as well. Currently looking for an elegant way to incrementally increase MY and it's a bit of a pain to try to keep it clean and consistent with the other code utilizing Lubridate.

zachmeador on 13 Jun 2019

This is a really common use case at our work and would be a great add =)

adanvr on 6 Jul 2019

+1 in favor of this feature request

Gillenwaters on 20 Jul 2019

Seems like floor_date covers most use-cases for this. For example,

dates <- as.Date(c("2019-01-01", "2019-02-20"))
floor_date(dates, "month")

Admittedly you get dates on the first of each month and not dates with no day component, but from that point on you can use all the standard Lubridate functions to add or subtract months.