add exact EBM test for regression, and multiclass

Author: paulbkoch

python/interpret-core/interpret-core.pyproj

@@ -117,6 +117,7 @@
     <Compile Include="tests\glassbox\ebm\research\__init__.py" />
     <Compile Include="tests\glassbox\ebm\test_bin.py" />
     <Compile Include="tests\glassbox\ebm\test_ebm.py" />
+    <Compile Include="tests\glassbox\ebm\test_ebm_exact.py" />
     <Compile Include="tests\glassbox\ebm\test_ebm_utils.py" />
     <Compile Include="tests\glassbox\ebm\test_merge_ebms.py" />
     <Compile Include="tests\glassbox\ebm\test_multiclass.py" />

python/interpret-core/tests/glassbox/ebm/test_ebm.py

@@ -1243,34 +1243,3 @@ def test_replicatability_classification():
         if total1 != total2:
             assert total1 == total2
             break
-
-
-def test_identical_classification():
-    from interpret.develop import get_option, set_option
-
-    original = get_option("acceleration")
-    set_option("acceleration", 0)
-
-    for iteration in range(1):
-        total = 0.0
-        seed = 0
-        for i in range(10):
-            X, y, names, types = make_synthetic(
-                seed=seed, classes=2, output_type="float", n_samples=250
-            )
-            seed += 1
-
-            ebm = ExplainableBoostingClassifier(
-                names, types, random_state=seed, max_rounds=10
-            )
-            ebm.fit(X, y)
-
-            pred = ebm.eval_terms(X)
-            total += np.sum(pred)
-
-        expected = -3.941291737419306e-15
-        if total != expected:
-            assert total == expected
-            break
-
-    set_option("acceleration", original)

python/interpret-core/tests/glassbox/ebm/test_ebm_exact.py

@@ -0,0 +1,48 @@
+# Copyright (c) 2023 The InterpretML Contributors
+# Distributed under the MIT software license
+
+from interpret.glassbox import (
+    ExplainableBoostingClassifier,
+    ExplainableBoostingRegressor,
+)
+from interpret.utils import make_synthetic
+from interpret.develop import get_option, set_option
+from interpret.utils._native import Native
+
+
+def test_identical_classification():
+    original = get_option("acceleration")
+    set_option("acceleration", 0)
+
+    total = 0.0
+    seed = 0
+    for n_classes in range(Native.Task_Regression, 4):
+        if n_classes < 2 and n_classes != Native.Task_Regression:
+            continue
+
+        classes = None if n_classes == Native.Task_Regression else n_classes
+
+        for iteration in range(2):
+            X, y, names, types = make_synthetic(
+                seed=seed, classes=classes, output_type="float", n_samples=257
+            )
+
+            ebm_type = (
+                ExplainableBoostingClassifier
+                if 0 <= n_classes
+                else ExplainableBoostingRegressor
+            )
+            ebm = ebm_type(names, types, random_state=seed, max_rounds=10)
+            ebm.fit(X, y)
+
+            pred = ebm._predict_score(X)
+            total += sum(pred.flat)
+
+            seed += 1
+
+    expected = 604.4169336846871
+
+    if total != expected:
+        assert total == expected
+
+    set_option("acceleration", original)