R/lubridate-pacea-series.R

In pbs-assess/PACea: An R package of Pacific ecosystem information to help facilitate an ecosystem approach to fisheries management

##' Convert an object that represents a time seris (`pacea_index` etc.) to have lubridate
##'  year/month column, and smooth over a year if required
##'
##' Called from `plot.pacea_index()` and others to give a lubrdiate column for
##' easier plotting.
##'
##' @inherit plot.pacea_index
##' @return original object with month values smoothed (averaged) over the year if
##'   requested, and a new `date` column in lubridate format. Sets a resulting
##'   date to be the first of the month or Jan 1st of that year (as appropriate),
##'   for easy
##'   plotting, but note that these should not be taken as exact dates, which is
##'   why we saved the original objects just as year and/or month, not lubridate
##'   dates.
##' @export
##' @author Andrew Edwards
##' @examples
##' \dontrun{
##' lubridate_pacea_series(oni)
##' lubridate_pacea_series(oni, smooth_over_year=TRUE)
##'
##' }
lubridate_pacea_series <- function(obj,
                                   smooth_over_year = FALSE){

  if(smooth_over_year){
    stopifnot("to smooth over year you need monthly data (if you have daily we can adapt the code
               to use that); set smooth_over_year = FALSE" =
              "month" %in% names(obj))

    obj_lub <- dplyr::group_by(obj,
                               year) %>%
      dplyr::summarise(across(-month,
                              mean))  # Replace val, anom, and any other
                                              #  non-year non-month column with their
                                              #  annual mean

    obj_lub <- dplyr::mutate(obj_lub,
                             date = lubridate::ymd(year,
                                                   truncated = 2))
                             # sets date to 1st Jan of that year to give a valid
                             #  date; could change to middle of year, but a
                             #  little confusing. year column still retained
                             #  (but object not returned so okay).
  } else {
    if("month" %in% names(obj)){

      # TODO extract date-related columns automatically and create the date column correctly
      #  This works for oni, may need a switch (or function, since may want for
      #  pacea_st also) for years-only. And if make function then use for the
      #  obj_lub line above also.

      obj_lub <- dplyr::mutate(obj,
                               date = paste(year,
                                            month,
                                            sep = "-"))
      obj_lub$date <- lubridate::ym(obj_lub$date)

    } else {
    obj_lub <- dplyr::mutate(obj,
                             date = lubridate::ymd(year,
                                                   truncated = 2))
    }
  }
  return(obj_lub)
}