Counting Replacements with str_replace_all in a Dplyr Workflow

As a data analyst, it’s not uncommon to encounter messy data frames that require cleaning and preprocessing. One common task is replacing typos or incorrect values with correct ones. In this article, we’ll explore how to count the number of replacements made by str_replace_all in a dplyr workflow.

Introduction

The dplyr package provides an efficient way to manipulate data frames using verbs like mutate, select, and arrange. However, when dealing with typos or incorrect values, finding all instances of the typo can be time-consuming, especially for large data frames. In this article, we’ll show how to count the number of replacements made by str_replace_all in a dplyr workflow.

Understanding str_replace_all

The str_replace_all function replaces specified values with new ones in a string vector or character column of a data frame. It returns a new string vector with all occurrences replaced.

Syntax

str_replace_all(x, pattern, replacement)

• x: The input string vector or character column.
• pattern: A character vector containing the values to be replaced.
• replacement: A single value that will replace each occurrence of pattern.

Counting Replacements

To count the number of replacements made by str_replace_all, we need to create a named vector for counting the occurrences of our typos and then map through each of them.

Example Code

library(tidyverse)

my_df <- data.frame(my_str = c("a", "ca", "c", "bla"))

my_typo_corrections <- c("a" = "b",
                         "c" = "d",
                         "what so ever" = "whatever")

typo_counts <- names(my_typo_corrections)

names(typo_counts) <- typo_counts

my_df |>
  mutate(my_str_new = str_replace_all(my_str, my_typo_corrections)) |> 
  rowwise() |> 
  mutate(counts = list(map(!!typo_counts,
                          ~str_count(my_str, .x)))) |> 
  ungroup() |> 
  unnest_wider(counts)

This code creates a named vector typo_counts containing the names of our typos. We then use rowwise() to iterate over each row in the data frame and calculate the count of occurrences for each typo using str_count(). The result is a new column counts with the number of replacements made for each defined typo.

Result

The output will be:

# A tibble: 4 × 5
  my_str my_str_new     a     c `what so ever`
       <chr>          <chr>  <int> <int>          <int>
1 a            b             3     0              0
2 ca           db             2     2              0
3 c             d             0     2              0
4 bla           blb            1     0              0

From here, we can build our table with the number of replacements made for each defined typo.

Conclusion

Counting replacements made by str_replace_all in a dplyr workflow is a straightforward process that requires creating a named vector for counting occurrences and then mapping through each typo. This approach saves time and effort compared to finding instances manually, especially for large data frames. By using this technique, you can efficiently clean your data and build accurate tables with the number of replacements made.

Last modified on 2023-05-23