using tidyr functions with collapse

[manalkamal]

Hi,
I am using complete/ fill functions from {tidyr} to a grouped dataframe to fill in missing dates and values. I would like to know if {collapse} functions have a workaround for tidyr? currently, it seems not; when I grouped the data using fgroup_by() then using complete(), fill(), it says "the groups attribute must be a data frame"..
Thanks.

[SebKrantz]

Thanks. Please tag me for a fast response. You should be able to use na_locf() or na_focb() in a grouped fmutate() call no?

[manalkamal]

@SebKrantz Thanks! I wanted to complete a sequence of dates first and then fill in the values with last observation. complete from tidyr did not work on a grouped call. so is there an alternative to the following steps to run over a grouped dataframe:

1. complete sequence of dates
2. fill in values - this is clear now with na_locf()

[SebKrantz]

@manalkamal please provide an example for a more specific answer. In general, yes, many things can be done using collapse or fastverse functions. Haven't felt the need to use tidyr in a while.

[manalkamal]

@SebKrantz the same example as in here:
https://stackoverflow.com/questions/66495265/refill-missing-dateyear-month-and-value-by-groups-in-r

What could be the alternative in {collapse} & {fastverse} for the complete() function - last line in the accepted answer?
Thanks!

[SebKrantz]

@manalkamal how about:

df %>%
  fmutate(yr_month = as.yearmon(yr_month)) %>%
  join(expand.grid(clinic = unique(.$clinic),
                   yr_month = seq(min(.$yr_month), max(.$yr_month), by = 1 / 12)),
       how = "full") %>%
  replace_na(0, cols = "rate") %>%
  roworder(clinic, yr_month)

By groups would be more difficult - need to change the expand.grid() accordingly.

[manalkamal]

@SebKrantz Super helpful. Thank you!!