[backport] [CI] Skip pyspark sparse tests. (dmlc#8675) (dmlc#8678)

Author: trivialfis

tests/ci_build/conda_env/cpu_test.yml

@@ -36,7 +36,8 @@ dependencies:
 - cloudpickle
 - shap
 - modin
+# TODO: Replace it with pyspark>=3.4 once 3.4 released.
+# - https://ml-team-public-read.s3.us-west-2.amazonaws.com/pyspark-3.4.0.dev0.tar.gz
+- pyspark>=3.3.1
 - pip:
   - datatable
-  # TODO: Replace it with pyspark>=3.4 once 3.4 released.
-  - https://ml-team-public-read.s3.us-west-2.amazonaws.com/pyspark-3.4.0.dev0.tar.gz

tests/python/test_spark/test_spark_local.py

@@ -41,6 +41,16 @@
 pytestmark = testing.timeout(60)
 
 
+def no_sparse_unwrap():
+    try:
+        from pyspark.sql.functions import unwrap_udt
+
+    except ImportError:
+        return {"reason": "PySpark<3.4", "condition": True}
+
+    return {"reason": "PySpark<3.4", "condition": False}
+
+
 class XgboostLocalTest(SparkTestCase):
     def setUp(self):
         logging.getLogger().setLevel("INFO")
@@ -985,6 +995,7 @@ def test_classifier_with_feature_names_types_weights(self):
         model = classifier.fit(self.cls_df_train)
         model.transform(self.cls_df_test).collect()
 
+    @pytest.mark.skipif(**no_sparse_unwrap())
     def test_regressor_with_sparse_optim(self):
         regressor = SparkXGBRegressor(missing=0.0)
         model = regressor.fit(self.reg_df_sparse_train)
@@ -1001,6 +1012,7 @@ def test_regressor_with_sparse_optim(self):
         for row1, row2 in zip(pred_result, pred_result2):
             self.assertTrue(np.isclose(row1.prediction, row2.prediction, atol=1e-3))
 
+    @pytest.mark.skipif(**no_sparse_unwrap())
     def test_classifier_with_sparse_optim(self):
         cls = SparkXGBClassifier(missing=0.0)
         model = cls.fit(self.cls_df_sparse_train)