pecan
pecan copied to clipboard
PecanProject
Replace `transform()` with something programming-friendly
Actually, my advice would be to save this for a separate PR. I did a quick search and there are many instances of
transform()in PEcAn which will all likely result in a similar problem of undefined global variables. It would probably easier to change them all at once when you've got the fix pattern figured out.
Originally posted by @Aariq in https://github.com/PecanProject/pecan/pull/2956#discussion_r921198756
transform() is similar to dplyr::mutate() but doesn't have a way of avoiding the 'undefined global variables' CRAN check error. The help file for transform() explicitly says to not use it in programming.
For example, transform(iris, Sepal.Length = Sepal.Length * 2) will result in an undefined global variable Sepal.Length. With dplyr's .data pronoun this can be avoided: dplyr::mutate(iris, Sepal.Length = .data$Sepal.Length * 2)
Hi @Aariq, I am not able to find any replacement for transform(). I think this Is the only function for changing data frames. What do you think?
Most uses of transform can be rewritten (sometimes much more wordily!) as explicit assignments to column names inside an explicitly named dataframe: new_data <- transform(data_with_tediously_long_name, some_column = some_other_column * 2) => new_data <- data_with_tediously_long_name; new_data$some_column <- new_data$some_other_column * 2
new_df <- transform(old_df, new_col = old_col *2)
If a package already imports dplyr then you can do:
new_df <- dplyr::mutate(old_df, new_col = .data$old_col *2)
otherwise you'd have to do something like:
new_df <- old_df
new_df$new_col <- new_df$old_col * 2