Isocalenderweek/year out of date to create a timeseries

[lherming]

Hello, everyone,

I have to make a time series on iso-calendar weeks. My data are given on a daily basis. I want to aggregate them on ISO calendar week level over the respective product groups. However, I want to have the calendar week and the year in the data and format this as date or POSIXct so that I can create a time series afterwards.
Summarized:
Date (d-m-y) -> Isoweek/year as a new column but formated as.date --> Create time series (frequency 52) per product group (both individually and one large one for all)

My previous attempts have not worked:
data2 <- data %>%
mutate(new_date=as.Date(KW, format="%W"), new_date2=year(Date))

data3 <- data %>%
mutate(new_date=isoweek(date), new_date2=year(date))

Thank you very much for your help.
Enclosed you will find an excerpt from my data so that you can understand the structure.

Kind regards,
Lukas

[technocrat]

lubridate::isoweek() returns the week as it would appear in the ISO 8601 system, which uses a reoccurring leap week.

Approach

your_data %>% mutate(iso_week = lubridate::isoweek(lubridate::dmy(Date)) -> output

[lherming]

Thank you for your fast reply.
But if I take your approach I get:
All formats failed to parse. No formats found.

And there are no values in the data (NA).

[technocrat]

See the help for lubridate.

dmy(15/07/2020)
ymd(2020-07-15)

and variations are available.

[lherming]

Fixed it thanks.
But it still only gives me the isoweek. It needs to include the year of the week as well. And its format is still num whereas I need date or POSIXct to make a ts() out of it. I am sorry about that many questions. Do you have a solution for that as well?

[technocrat]

Two issues then:

1. Trivially: extract isoweek and year from the datetime object and convert to a string
2. Possibly infeasible: converting that string to a datetime object

Is 2 required given 1?

[lherming]

It actually is cause I need to make a ts() out of this in the end.

[technocrat]

No worries, then. ts doesn't care. You just give it an argument for the start and frequency.

[lherming]

well it doesnt give me a two digit isoweek

[technocrat]

isoweek returns class num To left pad to make two-digit character strings:

date_string <- "2020-07-15"
date_object <- lubridate::ymd(date_string)
lubridate::isoweek(date_string)
#> [1] 29
date_string <- "2020-01-15"
date_object <- lubridate::ymd(date_string)
lubridate::isoweek(date_string)
#> [1] 3
stringr::str_pad(lubridate::isoweek(date_string),2,side = "left", pad = "0")
#> [1] "03"

^{Created on 2020-07-15 by the reprex package (v0.3.0)}