Merge pull request #1562 from exploratory-io/codex/issue-37226-r

Add adjusted chi-square residual output

Author: hidekoji

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: exploratory
 Type: Package
 Title: R package for Exploratory
-Version: 16.0.12
-Date: 2026-07-23
+Version: 16.0.13
+Date: 2026-07-24
 Authors@R: c(person("Hideaki", "Hayashi", email = "hideaki@exploratory.io", role = c("aut", "cre")), person("Hide", "Kojima", email = "hide@exploratory.io", role = c("aut")), person("Kan", "Nishida", email = "kan@exploratory.io", role = c("aut")), person("Kei", "Saito", email = "kei@exploratory.io", role = c("aut")), person("Yosuke", "Yasuda", email = "double.y.919.quick@gmail.com", role = c("aut")))
 URL: https://github.com/exploratory-io/exploratory_func
 Description: Functions for Exploratory

R/test_wrapper.R

@@ -534,6 +534,12 @@ tidy.chisq_exploratory <- function(x, type = "observed") {
     resid_df <- resid_df %>% tibble::rownames_to_column(var = x$var1)
     resid_df <- resid_df %>% tidyr::gather(!!rlang::sym(x$var2), "residual", -!!rlang::sym(x$var1))
 
+    # Keep the cell statistic and multiplicity correction aligned with the
+    # Correspondence Analysis report.
+    adjusted_resid_df <- as.data.frame(x$stdres)
+    adjusted_resid_df <- adjusted_resid_df %>% tibble::rownames_to_column(var = x$var1)
+    adjusted_resid_df <- adjusted_resid_df %>% tidyr::gather(!!rlang::sym(x$var2), "adjusted_standardized_residual", -!!rlang::sym(x$var1))
+
     resid_raw_df <- as.data.frame(x$observed - x$expected) # x$residual is standardized, but here, take raw difference between observed and expected. 
     resid_raw_df <- resid_raw_df %>% tibble::rownames_to_column(var = x$var1)
     resid_raw_df <- resid_raw_df %>% tidyr::gather(!!rlang::sym(x$var2), "residual_raw", -!!rlang::sym(x$var1))
@@ -544,6 +550,7 @@ tidy.chisq_exploratory <- function(x, type = "observed") {
 
     ret <- obs_df %>% left_join(expected_df, by=c(x$var1, x$var2)) # join expected column 
     ret <- ret %>% left_join(resid_df, by=c(x$var1, x$var2)) # join expected column
+    ret <- ret %>% left_join(adjusted_resid_df, by=c(x$var1, x$var2))
     ret <- ret %>% left_join(resid_raw_df, by=c(x$var1, x$var2)) # join residual_raw column
     ret <- ret %>% left_join(resid_ratio_df, by=c(x$var1, x$var2)) # join residual_ratio column
     if (is.nan(x$statistic) || x$statistic <= 0) {
@@ -552,6 +559,11 @@ tidy.chisq_exploratory <- function(x, type = "observed") {
     else {
       ret <- ret %>% mutate(contrib = 100*residual^2/(!!(x$statistic))) # add percent contribution too.
     }
+    ret <- ret %>%
+      mutate(
+        adjusted_p_value = 2 * stats::pnorm(abs(adjusted_standardized_residual), lower.tail = FALSE),
+        adjusted_p_value = stats::p.adjust(adjusted_p_value, method = "holm")
+      )
 
     if (!is.null(x$var1_levels)) {
       ret[[x$var1]] <- factor(ret[[x$var1]], levels=x$var1_levels)

exploratory_16.0.8.tgz

@@ -291,6 +291,17 @@ test_that("test exp_chisq", {
   observed <- ret %>% tidy_rowwise(model, type="observed")
   summary <- ret %>% glance_rowwise(model)
   residuals <- ret %>% tidy_rowwise(model, type="residuals")
+  expect_true(all(c("adjusted_standardized_residual", "adjusted_p_value") %in% colnames(residuals)))
+  expect_equal(
+    residuals$adjusted_standardized_residual,
+    as.vector(ret$model[[1]]$stdres),
+    tolerance = 1e-12
+  )
+  expect_equal(
+    residuals$adjusted_p_value,
+    p.adjust(2 * pnorm(abs(residuals$adjusted_standardized_residual), lower.tail = FALSE), method = "holm"),
+    tolerance = 1e-12
+  )
   expect_true(all(c("Cramer's V","Chi-Square","DF","P Value","Cohen's W",
                     "Power", "Type 2 Error","Rows") %in% colnames(summary)
   ))
@@ -1085,4 +1096,4 @@ test_that("1-way ANOVA with complex column names works for type='emmeans' and ty
   expect_true(is.data.frame(ret_emmeans))
   ret_pairs <- model_df %>% tidy_rowwise(model, type = "pairs", pairs_adjust = "tukey")
   expect_true(is.data.frame(ret_pairs))
-})
+})