R

Package Development
Misc Information

Package Creation

Testing

Documenting

Tips

Tidy Expressions vs. Standard Evaluation

Tidy Expressions

Tidy expressions are used in functions that leverage tidyselect syntax, like select(), pivot_wider(), and rename(). These functions support special selection helpers like starts_with(), ends_with(), and others. They don’t work well with .data because they expect column names as strings or with selection helpers.

Standard Evaluation

Standard evaluation refers to functions that evaluate arguments directly without special tidyselect syntax. Functions like filter() and mutate() work well with .data$variable because they evaluate expressions directly instead of converting them to strings.

Summary

Use .data$variable for standard evaluation functions like filter().

Use dplyr::all_of() for functions using tidyselect expressions like pivot_wider().

Functions & Quasiquotation ({{}}, !!, :=)

When writing functions that accept column names as arguments, tidy evaluation tools come into play:

{{}} (Curly-curly): Used within functions to embrace a column name. It is shorthand for enquo() and !!.

!! (Bang-bang): Unquotes a captured expression, making it evaluated within the context of the data.

:= (Colon-equals): Often used with dplyr::mutate() and dplyr::summarize() to dynamically assign names to columns.

Example

summarize_and_mutate <- function(data, group_var, stat_var, new_var_name, multiplier = 1) {
    data %>%
        dplyr::group_by({{ group_var }}) %>%
        dplyr::summarize(mean_value = mean({{ stat_var }}, na.rm = TRUE)) %>%
        dplyr::mutate("{new_var_name}" := mean_value * !!multiplier)
}

Common Errors

1:

no visible binding for global variable ‘stat_name’

dplyr::filter(stat_name == "N")

solution:

dplyr::filter(.data$stat_name == "N")

no visible binding for global variable ‘variable’ no visible binding for global variable ‘stat’

tidyr::pivot_wider(names_from = variable, values_from = stat)

our initial insinct is to use the following after what we just learned but this wont work because of pivot_wider is a tidy select expression like select and arrange

tidyr::pivot_wider(names_from = .data$variable, values_from = .data$stat)

# Use of .data in tidyselect expressions was deprecated in tidyselect 1.2.0.
# i Please use `"stat"` instead of `.data$stat`