在R中对Iris数据集列循环执行Dunn检验时列名获取失败的问题排查

Hey there! Let's break down why your code isn't running properly and get it working smoothly.

The Root Cause #

The main issue lies in how you're referencing column names within the dunn_test() formula. Using get() inside the formula syntax doesn't play nicely with rstatix's tidyeval-based interface. The dunn_test() function expects formula terms it can resolve directly in the dataset, and get() fails to translate string column names into valid formula terms correctly in this context.

Fixed Code Options #

Here are two working versions of your code, both addressing the formula reference problem:

Option 1: Using Tidyeval (sym() and !!)

This converts your string column names into symbols that the formula can interpret correctly:

library(rstatix)
data <- iris

# Predefine the group column name since it's always Species (no need to re-fetch it in the loop)
group_col <- colnames(data)[5]

for (i in seq(1:4)) {
  feature_col <- colnames(data)[i]
  # Convert string to symbol and unquote it for the formula
  dtest <- data %>% 
    dunn_test( !!sym(feature_col) ~ !!sym(group_col), p.adjust.method = "BH")
  print(dtest)
  print(paste("Completed test for feature:", feature_col))
}

Option 2: Using reformulate()

This function builds a formula directly from strings, which is often more straightforward for dynamic formula scenarios:

library(rstatix)
data <- iris

group_col <- colnames(data)[5]

for (i in seq(1:4)) {
  feature_col <- colnames(data)[i]
  # Dynamically construct the test formula
  test_formula <- reformulate(response = feature_col, termlabels = group_col)
  dtest <- data %>% 
    dunn_test(formula = test_formula, p.adjust.method = "BH")
  print(dtest)
  print(paste("Completed test for feature:", feature_col))
}

Quick Optimization Note #

I moved the group column (Species) outside the loop since it never changes—this is just a small cleanup, not a fix for the core issue, but it makes the code a bit more efficient.

内容的提问来源于stack exchange，提问作者krishthw