Transform chr to date format in R - r

Question

I want to transform from chr to date format
I have this representing year -week:
2020-53
I ve tried to do this
mutate(semana=as_date(year_week,format="%Y-%U"))
but I get the same date in all dataset 2020-01-18
I also tried
mutate(semana=strptime(year_week, "%Y-%U"))
getting the same result
Here you can see the wrong convertion
Any idea?, thanks

Answer 1

I think I've got something that does the job.
library(tidyverse)
library(lubridate)
# Set up table like example in post
trybble <- tibble(year_week = c("2020-53", rep("2021-01", 5)),
country = c("UK", "FR", "GER", "ITA", "SPA", "UK"))
# Function to go into mutate with given info of year and week
y_wsetter <- function(fixme, yeargoal, weekgoal) {
lubridate::year(fixme) <- yeargoal
lubridate::week(fixme) <- weekgoal
return(fixme)
}
# Making a random date so col gets set right
rando <- make_datetime(year = 2021, month = 1, day = 1)
# Show time
trybble <- trybble %>%
add_column(semana = rando) %>% # Set up col of dates to fix
mutate(yerr = substr(year_week, 1, 4)) %>% # Get year as chr
mutate(week = substr(year_week, 6, 7)) %>% # Get week as chr
mutate(semana2 = y_wsetter(semana,
as.numeric(yerr),
as.numeric(week))) %>% # fixed dates
select(-c(yerr, week, semana))
Notes:
If you somehow plug in a week greater than 53, lubridate doesn't mind, and goes forward a year.
I really struggled to get mutate to play nicely without writing my own function y_wsetter. In my experience with mutates with multiple inputs, or where I'm changing a "property" of a value instead of the whole value itself, I need to probably write a function. I'm using the lubridate package to change just the year or week based on your year_week column, so this is one such situation where a quick function helps mutate out.
I was having a weird time when I tried setting rando to Sys.Date(), so I manually set it to something using make_datetime. YMMV