如何将含mutate与case_when的R代码封装为可复用函数？

Great question! Handling multi-option flag columns can get repetitive fast, so wrapping this logic into a reusable function is a smart move. Let's break down how to simplify those repeated case_when calls and build a flexible function that fits your needs.

Step 1: Understand the Repetitive Pattern #

First, let's recap the kind of code you're probably working with. For example, if you have a dataset with comma-separated interests, you might be writing code like this to create flag columns:

library(dplyr)
library(stringr)

sample_data <- tibble(
  id = 1:5,
  interests = c("读书,运动", "运动,音乐", "读书", "音乐,旅行", "读书,旅行,运动")
)

# Repetitive code you want to replace
sample_data %>%
  mutate(
    读书 = case_when(str_detect(interests, "读书") ~ 1, TRUE ~ 0),
    运动 = case_when(str_detect(interests, "运动") ~ 1, TRUE ~ 0),
    音乐 = case_when(str_detect(interests, "音乐") ~ 1, TRUE ~ 0),
    旅行 = case_when(str_detect(interests, "旅行") ~ 1, TRUE ~ 0)
  )

The core issue is repeating the same case_when(str_detect(...)) logic for every pattern. We can eliminate this repetition using purrr to batch-generate the flag columns.

Step 2: Build the Reusable Function #

Here's a flexible function that handles all your requirements: it accepts custom input data, target column, match patterns, option count (handled automatically by the length of your patterns), and output column names.

library(dplyr)
library(stringr)
library(purrr)

create_multi_flags <- function(data, target_col, patterns, output_names = NULL) {
  # Default to using patterns as output column names if none are provided
  if (is.null(output_names)) {
    output_names <- patterns
  }
  
  # Ensure pattern count matches output name count (prevents mismatches)
  stopifnot(length(patterns) == length(output_names), 
            "Number of patterns must match number of output column names")
  
  # Batch-generate flag columns using map_dfc
  flag_columns <- map_dfc(patterns, ~ {
    # For each pattern, detect presence in target column and convert to 1/0
    data %>%
      pull({{target_col}}) %>%
      str_detect(.x) %>%
      as.integer()
  }) %>%
    set_names(output_names)
  
  # Combine original data with new flag columns
  bind_cols(data, flag_columns)
}

Step 3: Use the Function #

Let's test this with your sample data. You can customize the output column names or let them default to the pattern strings:

# Sample data
sample_data <- tibble(
  id = 1:5,
  interests = c("读书,运动", "运动,音乐", "读书", "音乐,旅行", "读书,旅行,运动")
)

# Example 1: Custom output column names
result_with_custom_names <- create_multi_flags(
  data = sample_data,
  target_col = interests,
  patterns = c("读书", "运动", "音乐", "旅行"),
  output_names = c("flag_read", "flag_sport", "flag_music", "flag_travel")
)

# Example 2: Default output names (uses pattern strings)
result_with_default_names <- create_multi_flags(
  data = sample_data,
  target_col = interests,
  patterns = c("读书", "运动", "音乐", "旅行")
)

Expected Output #

For the first example with custom names, you'll get this clean output:

# A tibble: 5 × 6
     id interests         flag_read flag_sport flag_music flag_travel
  <int> <chr>                <int>      <int>      <int>       <int>
1     1 读书,运动                1          1          0           0
2     2 运动,音乐                0          1          1           0
3     3 读书                     1          0          0           0
4     4 音乐,旅行                0          0          1           1
5     5 读书,旅行,运动           1          1          0           1

Why This Works #

• No more repeated case_when: map_dfc iterates over each pattern and applies the same detection logic, so you only write the core logic once.
• Flexibility: You can pass any dataset, target column (using tidy evaluation with {{}}), patterns, and output names.
• Safety: The stopifnot check ensures you don't accidentally pass mismatched numbers of patterns and output names.

内容的提问来源于stack exchange，提问作者gofraidh