diff --git a/tests/unit-tests/gconstruct/test_transform.py b/tests/unit-tests/gconstruct/test_transform.py
index 54bed2d382..83cd7a16ae 100644
--- a/tests/unit-tests/gconstruct/test_transform.py
+++ b/tests/unit-tests/gconstruct/test_transform.py
@@ -722,6 +722,30 @@ def test_bucket_transform(out_dtype):
     feats_tar = np.array([[1, 1], [1, 1], [1, 1], [1, 1]], dtype=out_dtype)
     assert_equal(bucket_feats['test'], feats_tar)
 
+
+def test_multicolumn():
+    # Just get the features without transformation.
+    feat_op1 = [{
+        "feature_col": ["test1", "test2"],
+        "feature_name": "test3",
+    }]
+    (res, _, _) = parse_feat_ops(feat_op1)
+    assert len(res) == 1
+    assert res[0].col_name == feat_op1[0]["feature_col"]
+    assert res[0].feat_name == feat_op1[0]["feature_name"]
+    assert isinstance(res[0], Noop)
+
+    data = {
+        "test1": np.random.rand(4, 2),
+        "test2": np.random.rand(4, 2)
+    }
+    data["test3"] = np.column_stack((data['test1'], data['test2']))
+    proc_res = process_features(data, res)
+    assert "test3" in proc_res
+    assert proc_res["test3"].dtype == np.float32
+    np.testing.assert_allclose(proc_res["test3"], data["test3"])
+
+
 if __name__ == '__main__':
     test_categorize_transform()
     test_get_output_dtype()
@@ -745,3 +769,4 @@ def test_bucket_transform(out_dtype):
     test_collect_label_stats()
     test_custom_label_processor()
     test_classification_processor()
+    test_multicolumn()
\ No newline at end of file