Understanding Non-Standard Evaluation in R

=====================================================

Non-standard evaluation (NSE) is a feature of the R programming language that allows for more flexible and expressive syntax. In this answer, we will explore how to use NSE to achieve a specific goal.

Background

The original question provided a dataframe stage_refs with two columns new.diff.var and var.1 that were used as arguments in the difftime_fun function. The intention was to apply this function to each row of stage_refs, but the problem statement was encountering non-standard evaluation problems.

Solution Overview

The solution involves rewriting the difftime_fun function to accept a dataframe as an argument and using the .data pronoun to access the columns passed as character strings. We will then use purrr::reduce or base Reduce to loop over the rows of stage_refs and apply the difftime_fun to create new time difference columns.

Step-by-Step Solution

Step 1: Rewrite the Function

library(tidyverse)

difftime_fun <- function(.data, x, y, z) {
  .data |>
    mutate("{x}" := difftime(.data[[y]], .data[[z]]))
}

In this rewritten version of difftime_fun, we use the .data pronoun to access the dataframe passed as an argument. We then use mutate to create a new column with the time difference.

Step 2: Apply the Function

purrr::reduce(seq(nrow(stage_refs)), function(.x, i) {
  difftime_fun(.x,
    stage_refs$new.diff.var[[i]],
    stage_refs$var.1[[i]],
    stage_refs$var.2[[i]]
  )
}, .init = date_values)

Here, we use purrr::reduce to loop over the rows of stage_refs. In each iteration, we call difftime_fun with the current row’s values for new.diff.var, var.1, and var.2.

Step 3: Combine Results

The resulting code is wrapped in a single pipe (|>) to ensure that the output of each step is passed as input to the next.

Example Use Case

Let’s create a sample dataframe stage_refs with two columns new.diff.var and var.1, and then apply the solution:

# Create sample data
stage_refs <- data.frame(
  new.diff.var = c("time value 1", "time value 2", "time value 3"),
  var.1 = c("time value 4", "time value 5", "time value 6")
)

date_values <- data.frame()

# Define the function
difftime_fun <- function(.data, x, y, z) {
  .data |>
    mutate("{x}" := difftime(.data[[y]], .data[[z]]))
}

# Apply the solution
purrr::reduce(seq(nrow(stage_refs)), function(.x, i) {
  difftime_fun(.x,
    stage_refs$new.diff.var[[i]],
    stage_refs$var.1[[i]],
    stage_refs$var.2[[i]]
  )
}, .init = date_values)

# Print the results
print(date_values)

This code will create a new dataframe date_values with additional columns representing time differences.

Conclusion

In this answer, we explored how to use non-standard evaluation in R to achieve a specific goal. By rewriting the difftime_fun function to accept a dataframe as an argument and using the .data pronoun, we can easily apply this function to each row of another dataframe. We then used purrr::reduce or base Reduce to loop over the rows and create new time difference columns. This approach provides a flexible and expressive way to perform complex data transformations in R.

Last modified on 2025-04-29