Merge pull request #23 from dandls/dev_issue_22

Early stop for parallel plot if there are no numeric features. Fixes #22

R/Counterfactuals.R

@@ -278,6 +278,10 @@ Counterfactuals = R6::R6Class("Counterfactuals",
 
       is_numeric_col = sapply(dt, function(x) is.numeric(x))
       numeric_cols = names(dt)[is_numeric_col]
+      if (length(numeric_cols) == 0L) {
+        stop("Can only consider numeric features for parallel plot, but no numeric features present in data")
+      }
+
       non_numeric_cols = names(dt)[!is_numeric_col]
       if (length(non_numeric_cols) > 0L) {
         dt[, (non_numeric_cols) := NULL]

tests/testthat/_snaps/Counterfactuals.md

@@ -6,6 +6,10 @@
 
     Assertion on 'feature_names' failed: Names must be a subset of {'var_num_1','var_num_2','var_fact_1','var_fact_2'}, but has additional elements {'non_in_data1','non_in_data2'}.
 
+# plot_parallel returns error for if there are no numeric features
+
+    Can only consider numeric features for parallel plot, but no numeric features present in data
+
 # methods that require at least one counterfactuals are blocked when no counterfactuals found
 
     Assertion on 'self$data' failed: Must have at least 1 rows, but has 0 rows.

tests/testthat/test-Counterfactuals.R

@@ -122,6 +122,21 @@ test_that("plot_parallel returns error for unknown feature names", {
   expect_snapshot_error(cf$plot_surface(c("non_in_data1", "non_in_data2")))
 })
 
+test_that("plot_parallel returns error for if there are no numeric features", {
+  set.seed(1234)
+  df <- data.frame(
+    "a" = as.factor(sample(c("x", "y", "z"), size = 30, replace = TRUE)),
+    "b" = as.factor(sample(c("x", "y", "z"), size = 30, replace = TRUE)),
+    "c" = as.factor(sample(c("x", "y", "z"), size = 30, replace = TRUE)),
+    "target" = as.factor(sample(c("p", "n"), size = 30, replace = TRUE))
+  )
+  rf = randomForest(target ~ ., data = df, ntree = 5L)
+  pred = Predictor$new(rf, data = df, type = "class")
+  x_interest = head(subset(df, select = -target), n = 1L)
+  wi = WhatIfClassif$new(pred, n_counterfactuals = 5)
+  cfactuals = wi$find_counterfactuals(x_interest, desired_class = "p")
+  expect_snapshot_error(cfactuals$plot_parallel())
+})
 
 # $evaluate() ----------------------------------------------------------------------------------------------------------
 test_that("evaluate returns error if measures are not known", {