speeding up expandYearlyCosts

[TellAnAx]

Hi there,
when working with the invacost package, I noticed that the expandYearlyCosts() function is very slow. I ran a benchmark and processing the data according to the example you provide takes about 45 s on my machine.

From what I can see, the reason is that the following code is inefficient:

return(
    dplyr::bind_rows(
      lapply(costdb$Cost_ID, function(x, costdb.,
                                      start,
                                      end) { 
        years <- costdb.[which(costdb.$Cost_ID == x), start]:
          costdb.[which(costdb.$Cost_ID == x), end]
        return(data.frame(Impact_year = years,
                          costdb.[which(costdb.$Cost_ID == x), ][
                            rep(seq_len(nrow(costdb.[costdb.$Cost_ID == x, ])), 
                                each = length(years)), ]))
      }, costdb. = costdb, start = startcolumn, end = endcolumn)
    )
  )

The costdb. is subsetted by Cost_ID, then expanded, the resulting dataframe saved as element of a list and then the entire list is merged rowwise. This process could be done much more efficiently by using some tidyverse functions. I would suggest an update. 😃

[Farewe]

Hi @TellAnAx ,

thanks for your feedback. That's indeed an issue - not too much of an issue because the database is not that big, but still an improvement. I coded that before AI was there to help us optimize the code!

I have improved the code for this function in another, unpublished project, so I can implement this change when I will release a new version of the package (the next version will ship a currency conversion function). Hopefully, before the end of the year - I am waiting for a colleague's help on the currency conversion function.

[TellAnAx]

Hi @Farewe,

yes, I also see it as a simple enhancement that will reduce computing cost. Looking forward to the new version!