chore: fix that sklearn will be imported when vizer hptuning is not used

PiperOrigin-RevId: 566442578

Author: jaycee-li

vertexai/preview/hyperparameter_tuning/vizier_hyperparameter_tuner.py

@@ -69,42 +69,32 @@
 _RMSLE_METRIC_ID = "rmsle"
 _MSE_METRIC_ID = "mse"
 
-try:  # Only used by local tuning loop
-    import sklearn.metrics
-    from sklearn.model_selection import train_test_split
-
-    _SUPPORTED_METRIC_FUNCTIONS = {
-        _ROC_AUC_METRIC_ID: sklearn.metrics.roc_auc_score,
-        _F1_METRIC_ID: sklearn.metrics.f1_score,
-        _PRECISION_METRIC_ID: sklearn.metrics.precision_score,
-        _RECALL_METRIC_ID: sklearn.metrics.recall_score,
-        _ACCURACY_METRIC_ID: sklearn.metrics.accuracy_score,
-        _MAE_METRIC_ID: sklearn.metrics.mean_absolute_error,
-        _MAPE_METRIC_ID: sklearn.metrics.mean_absolute_percentage_error,
-        _R2_METRIC_ID: sklearn.metrics.r2_score,
-        _RMSE_METRIC_ID: functools.partial(
-            sklearn.metrics.mean_squared_error, squared=False
-        ),
-        _RMSLE_METRIC_ID: functools.partial(
-            sklearn.metrics.mean_squared_log_error, squared=False
-        ),
-        _MSE_METRIC_ID: sklearn.metrics.mean_squared_error,
-    }
-    _SUPPORTED_METRIC_IDS = frozenset(_SUPPORTED_METRIC_FUNCTIONS.keys()).union(
-        frozenset([_CUSTOM_METRIC_ID])
-    )
-    _SUPPORTED_CLASSIFICATION_METRIC_IDS = frozenset(
-        [
-            _ROC_AUC_METRIC_ID,
-            _F1_METRIC_ID,
-            _PRECISION_METRIC_ID,
-            _RECALL_METRIC_ID,
-            _ACCURACY_METRIC_ID,
-        ]
-    )
+_SUPPORTED_METRIC_IDS = frozenset(
+    [
+        _CUSTOM_METRIC_ID,
+        _ROC_AUC_METRIC_ID,
+        _F1_METRIC_ID,
+        _PRECISION_METRIC_ID,
+        _RECALL_METRIC_ID,
+        _ACCURACY_METRIC_ID,
+        _MAE_METRIC_ID,
+        _MAPE_METRIC_ID,
+        _R2_METRIC_ID,
+        _RMSE_METRIC_ID,
+        _RMSLE_METRIC_ID,
+        _MSE_METRIC_ID,
+    ]
+)
+_SUPPORTED_CLASSIFICATION_METRIC_IDS = frozenset(
+    [
+        _ROC_AUC_METRIC_ID,
+        _F1_METRIC_ID,
+        _PRECISION_METRIC_ID,
+        _RECALL_METRIC_ID,
+        _ACCURACY_METRIC_ID,
+    ]
+)
 
-except ImportError:
-    pass
 
 # Vizier client constnats
 _STUDY_NAME_PREFIX = "vizier_hyperparameter_tuner_study"
@@ -366,6 +356,13 @@ def _create_train_and_test_splits(
                 "test_fraction must be greater than 0 and less than 1 but was "
                 f"{test_fraction}."
             )
+        try:
+            from sklearn.model_selection import train_test_split
+        except ImportError:
+            raise ImportError(
+                "scikit-learn must be installed to create train and test splits. "
+                "Please call `pip install scikit-learn>=0.24`"
+            ) from None
 
         if isinstance(y, str):
             try:
@@ -414,6 +411,32 @@ def score(x_test, y_test):
         Returns:
             A tuple containing the model and the corresponding metric value.
         """
+        try:  # Only used by local tuning loop
+            import sklearn.metrics
+
+            _SUPPORTED_METRIC_FUNCTIONS = {
+                _ROC_AUC_METRIC_ID: sklearn.metrics.roc_auc_score,
+                _F1_METRIC_ID: sklearn.metrics.f1_score,
+                _PRECISION_METRIC_ID: sklearn.metrics.precision_score,
+                _RECALL_METRIC_ID: sklearn.metrics.recall_score,
+                _ACCURACY_METRIC_ID: sklearn.metrics.accuracy_score,
+                _MAE_METRIC_ID: sklearn.metrics.mean_absolute_error,
+                _MAPE_METRIC_ID: sklearn.metrics.mean_absolute_percentage_error,
+                _R2_METRIC_ID: sklearn.metrics.r2_score,
+                _RMSE_METRIC_ID: functools.partial(
+                    sklearn.metrics.mean_squared_error, squared=False
+                ),
+                _RMSLE_METRIC_ID: functools.partial(
+                    sklearn.metrics.mean_squared_log_error, squared=False
+                ),
+                _MSE_METRIC_ID: sklearn.metrics.mean_squared_error,
+            }
+        except Exception as e:
+            raise ImportError(
+                "scikit-learn must be installed to evaluate models. "
+                "Please call `pip install scikit-learn>=0.24`"
+            ) from e
+
         if self.metric_id == _CUSTOM_METRIC_ID:
             metric_value = model.score(x_test, y_test)
         else: