From 5d7f12dfe9e575b02df30b0365a926d821b1e076 Mon Sep 17 00:00:00 2001
From: openhands <openhands@all-hands.dev>
Date: Wed, 13 Aug 2025 02:38:07 +0000
Subject: [PATCH] Fix Issue #6: Add support for single column data

- Remove requirement for at least 2 columns in varimpact()
- Add vector input support by converting to data frame
- Add warning when using single variable for analysis
- Fix character column conversion in separate_factors_numerics()
- Add comprehensive tests for single column scenarios
- Maintain backward compatibility with existing functionality

Addresses issue #6: 'Single column in data' where users reported
'Error: invalid type (list) of argument' when using single variables.
---
 R/separate_factors_numerics.R       |   7 +-
 R/varimpact.R                       |  13 +++-
 tests/testthat/test-single-column.R | 106 ++++++++++++++++++++++++++++
 3 files changed, 122 insertions(+), 4 deletions(-)
 create mode 100644 tests/testthat/test-single-column.R

diff --git a/R/separate_factors_numerics.R b/R/separate_factors_numerics.R
index 2c6b20b..53e00c1 100644
--- a/R/separate_factors_numerics.R
+++ b/R/separate_factors_numerics.R
@@ -19,7 +19,12 @@ separate_factors_numerics =
     is_char = sapply(data, is.character)
 
     # Convert strings to factors.
-    data[, is_char] = sapply(data[, is_char], as.factor)
+    if (any(is_char)) {
+      # Convert each character column to factor individually
+      for (col_name in names(data)[is_char]) {
+        data[[col_name]] = as.factor(data[[col_name]])
+      }
+    }
   }
 
   # Identify factors.
diff --git a/R/varimpact.R b/R/varimpact.R
index 1874d62..792d1dd 100644
--- a/R/varimpact.R
+++ b/R/varimpact.R
@@ -218,9 +218,16 @@ varimpact =
   ######################
   # Argument checks.
 
-  # Confirm that data has at least two columns.
-  if (ncol(data) < 2L) {
-    stop("Data argument must have at least two columns.")
+  # Handle vector input by converting to data frame
+  if (is.vector(data) && !is.list(data)) {
+    if (verbose) cat("Converting vector input to data frame.\n")
+    data <- data.frame(X1 = data)
+  }
+
+  # Handle single column case with warning
+  if (ncol(data) == 1L) {
+    warning("Using single variable for variable importance analysis. Results may be limited.")
+    if (verbose) cat("Single variable detected in data.\n")
   }
 
   # Ensure that Y is numeric; e.g. can't be a factor.
diff --git a/tests/testthat/test-single-column.R b/tests/testthat/test-single-column.R
new file mode 100644
index 0000000..bfb7d0f
--- /dev/null
+++ b/tests/testthat/test-single-column.R
@@ -0,0 +1,106 @@
+# Test single column support in varimpact
+# This addresses issue #6: "Single column in data"
+
+library(testthat)
+library(varimpact)
+
+context("Single column support")
+
+test_that("separate_factors_numerics handles single columns", {
+  # Test single numeric column
+  data_numeric <- data.frame(x1 = c(1.1, 2.2, 3.3, 4.4, 5.5))
+  result <- separate_factors_numerics(data_numeric)
+  expect_equal(ncol(result$df_factors), 0)
+  expect_equal(ncol(result$df_numerics), 1)
+  expect_equal(colnames(result$df_numerics), "x1")
+  
+  # Test single factor column
+  data_factor <- data.frame(x1 = factor(c("A", "B", "C", "A", "B")))
+  result <- separate_factors_numerics(data_factor)
+  expect_equal(ncol(result$df_factors), 1)
+  expect_equal(ncol(result$df_numerics), 0)
+  expect_equal(colnames(result$df_factors), "x1")
+  
+  # Test single character column (should be converted to factor)
+  data_char <- data.frame(x1 = c("A", "B", "C", "A", "B"), stringsAsFactors = FALSE)
+  result <- separate_factors_numerics(data_char)
+  expect_equal(ncol(result$df_factors), 1)
+  expect_equal(ncol(result$df_numerics), 0)
+  expect_equal(colnames(result$df_factors), "x1")
+  expect_true(is.factor(result$df_factors$x1))
+})
+
+test_that("varimpact handles vector input", {
+  # Create test data
+  set.seed(1)
+  N <- 50
+  X_vector <- rnorm(N)
+  Y <- rbinom(N, 1, plogis(0.2 * X_vector))
+  
+  # Test that vector input is converted to data frame and processed
+  expect_warning(
+    vim <- varimpact(Y = Y, data = X_vector, verbose = FALSE, V = 2L),
+    "Using single variable for variable importance analysis"
+  )
+  
+  # Check that the result is valid
+  expect_s3_class(vim, "varimpact")
+  expect_true(is.list(vim$results_all))
+})
+
+test_that("varimpact handles single column data frame", {
+  # Create test data
+  set.seed(1)
+  N <- 50
+  X_single <- data.frame(x1 = rnorm(N))
+  Y <- rbinom(N, 1, plogis(0.2 * X_single$x1))
+  
+  # Test that single column data frame is processed with warning
+  expect_warning(
+    vim <- varimpact(Y = Y, data = X_single, verbose = FALSE, V = 2L),
+    "Using single variable for variable importance analysis"
+  )
+  
+  # Check that the result is valid
+  expect_s3_class(vim, "varimpact")
+  expect_true(is.list(vim$results_all))
+})
+
+test_that("varimpact handles single factor column", {
+  # Create test data
+  set.seed(1)
+  N <- 50
+  X_factor <- data.frame(x1 = factor(sample(c("A", "B", "C"), N, replace = TRUE)))
+  Y <- rbinom(N, 1, plogis(0.2 * as.numeric(X_factor$x1)))
+  
+  # Test that single factor column is processed with warning
+  expect_warning(
+    vim <- varimpact(Y = Y, data = X_factor, verbose = FALSE, V = 2L),
+    "Using single variable for variable importance analysis"
+  )
+  
+  # Check that the result is valid
+  expect_s3_class(vim, "varimpact")
+  expect_true(is.list(vim$results_all))
+})
+
+test_that("varimpact still works with multiple columns (regression test)", {
+  # Create test data
+  set.seed(1)
+  N <- 50
+  X_multi <- data.frame(
+    x1 = rnorm(N),
+    x2 = rnorm(N),
+    x3 = factor(sample(c("A", "B"), N, replace = TRUE))
+  )
+  Y <- rbinom(N, 1, plogis(0.2 * X_multi$x1 + 0.1 * X_multi$x2))
+  
+  # Test that multiple columns work without warning
+  expect_silent(
+    vim <- varimpact(Y = Y, data = X_multi, verbose = FALSE, V = 2L)
+  )
+  
+  # Check that the result is valid
+  expect_s3_class(vim, "varimpact")
+  expect_true(is.list(vim$results_all))
+})
\ No newline at end of file