Remove old callback deprecated in 1.3. (#7280)

tests/python/test_basic.py

@@ -76,23 +76,6 @@ class TestBasic:
             predt_1 = booster.predict(dtrain)
             np.testing.assert_allclose(predt_0, predt_1)
 
-    def test_record_results(self):
-        dtrain = xgb.DMatrix(dpath + 'agaricus.txt.train')
-        dtest = xgb.DMatrix(dpath + 'agaricus.txt.test')
-        param = {'max_depth': 2, 'eta': 1, 'verbosity': 0,
-                 'objective': 'binary:logistic', 'eval_metric': 'error'}
-        # specify validations set to watch performance
-        watchlist = [(dtest, 'eval'), (dtrain, 'train')]
-        num_round = 2
-        result = {}
-        res2 = {}
-        xgb.train(param, dtrain, num_round, watchlist,
-                  callbacks=[xgb.callback.record_evaluation(result)])
-        xgb.train(param, dtrain, num_round, watchlist,
-                  evals_result=res2)
-        assert result['train']['error'][0] < 0.1
-        assert res2 == result
-
     def test_multiclass(self):
         dtrain = xgb.DMatrix(dpath + 'agaricus.txt.train')
         dtest = xgb.DMatrix(dpath + 'agaricus.txt.test')
@@ -254,8 +237,18 @@ class TestBasic:
         ]
 
         # Use callback to log the test labels in each fold
-        def cb(cbackenv):
-            print([fold.dtest.get_label() for fold in cbackenv.cvfolds])
+        class Callback(xgb.callback.TrainingCallback):
+            def __init__(self) -> None:
+                super().__init__()
+
+            def after_iteration(
+                self, model,
+                epoch: int,
+                evals_log: xgb.callback.TrainingCallback.EvalsLog
+            ):
+                print([fold.dtest.get_label() for fold in model.cvfolds])
+
+        cb = Callback()
 
         # Run cross validation and capture standard out to test callback result
         with tm.captured_output() as (out, err):

tests/python/test_callback.py

@@ -249,12 +249,9 @@ class TestCallbacks:
             assert booster.num_boosted_rounds() == \
                 booster.best_iteration + early_stopping_rounds + 1
 
-    def run_eta_decay(self, tree_method, deprecated_callback):
+    def run_eta_decay(self, tree_method):
         """Test learning rate scheduler, used by both CPU and GPU tests."""
-        if deprecated_callback:
-            scheduler = xgb.callback.reset_learning_rate
-        else:
-            scheduler = xgb.callback.LearningRateScheduler
+        scheduler = xgb.callback.LearningRateScheduler
 
         dpath = os.path.join(tm.PROJECT_ROOT, 'demo/data/')
         dtrain = xgb.DMatrix(dpath + 'agaricus.txt.train')
@@ -262,10 +259,7 @@ class TestCallbacks:
         watchlist = [(dtest, 'eval'), (dtrain, 'train')]
         num_round = 4
 
-        if deprecated_callback:
-            warning_check = pytest.warns(UserWarning)
-        else:
-            warning_check = tm.noop_context()
+        warning_check = tm.noop_context()
 
         # learning_rates as a list
         # init eta with 0 to check whether learning_rates work
@@ -339,19 +333,9 @@ class TestCallbacks:
         with warning_check:
             xgb.cv(param, dtrain, num_round, callbacks=[scheduler(eta_decay)])
 
-    @pytest.mark.parametrize(
-        "tree_method, deprecated_callback",
-        [
-            ("hist", True),
-            ("hist", False),
-            ("approx", True),
-            ("approx", False),
-            ("exact", True),
-            ("exact", False),
-        ],
-    )
-    def test_eta_decay(self, tree_method, deprecated_callback):
-        self.run_eta_decay(tree_method, deprecated_callback)
+    @pytest.mark.parametrize("tree_method", ["hist", "approx", "exact"])
+    def test_eta_decay(self, tree_method):
+        self.run_eta_decay(tree_method)
 
     def test_check_point(self):
         from sklearn.datasets import load_breast_cancer

tests/python/test_survival.py

@@ -22,15 +22,25 @@ def test_aft_survival_toy_data():
     # "Accuracy" = the number of data points whose ranged label (y_lower, y_upper) includes
     #              the corresponding predicted label (y_pred)
     acc_rec = []
-    def my_callback(env):
-        y_pred = env.model.predict(dmat)
-        acc = np.sum(np.logical_and(y_pred >= y_lower, y_pred <= y_upper)/len(X))
-        acc_rec.append(acc)
+
+    class Callback(xgb.callback.TrainingCallback):
+        def __init__(self):
+            super().__init__()
+
+        def after_iteration(
+            self, model: xgb.Booster,
+            epoch: int,
+            evals_log: xgb.callback.TrainingCallback.EvalsLog
+        ):
+            y_pred = model.predict(dmat)
+            acc = np.sum(np.logical_and(y_pred >= y_lower, y_pred <= y_upper)/len(X))
+            acc_rec.append(acc)
+            return False
 
     evals_result = {}
-    params = {'max_depth': 3, 'objective':'survival:aft', 'min_child_weight': 0}
+    params = {'max_depth': 3, 'objective': 'survival:aft', 'min_child_weight': 0}
     bst = xgb.train(params, dmat, 15, [(dmat, 'train')], evals_result=evals_result,
-                    callbacks=[my_callback])
+                    callbacks=[Callback()])
 
     nloglik_rec = evals_result['train']['aft-nloglik']
     # AFT metric (negative log likelihood) improve monotonically