workbench 0.8.198__py3-none-any.whl → 0.8.203__py3-none-any.whl

workbench/core/artifacts/endpoint_core.py

@@ -30,12 +30,14 @@ from sagemaker import Predictor
 
 # Workbench Imports
 from workbench.core.artifacts.artifact import Artifact
-from workbench.core.artifacts import FeatureSetCore, ModelCore, ModelType
+from workbench.core.artifacts import FeatureSetCore, ModelCore, ModelType, ModelFramework
 from workbench.utils.endpoint_metrics import EndpointMetrics
 from workbench.utils.cache import Cache
 from workbench.utils.s3_utils import compute_s3_object_hash
 from workbench.utils.model_utils import uq_metrics
-from workbench.utils.xgboost_model_utils import cross_fold_inference
+from workbench.utils.xgboost_model_utils import cross_fold_inference as xgboost_cross_fold
+from workbench.utils.pytorch_utils import pull_cv_results as pytorch_pull_cv
+from workbench.utils.chemprop_utils import pull_cv_results as chemprop_pull_cv
 from workbench_bridges.endpoints.fast_inference import fast_inference
 
 
@@ -387,7 +389,7 @@ class EndpointCore(Artifact):
         # Grab the model features and target column
         model = ModelCore(self.model_name)
         features = model.features()
-        target_column = model.target()
+        targets = model.target()  # Note: We have multi-target models (so this could be a list)
 
         # Run predictions on the evaluation data
         prediction_df = self._predict(eval_df, features, drop_error_rows)
@@ -395,45 +397,84 @@ class EndpointCore(Artifact):
             self.log.warning("No predictions were made. Returning empty DataFrame.")
             return prediction_df
 
+        # FIXME: Multi-target support - currently uses first target for metrics
+        # Normalize targets to handle both string and list formats
+        if isinstance(targets, list):
+            primary_target = targets[0] if targets else None
+        else:
+            primary_target = targets
+
         # Sanity Check that the target column is present
-        if target_column and (target_column not in prediction_df.columns):
-            self.log.important(f"Target Column {target_column} not found in prediction_df!")
+        if primary_target and (primary_target not in prediction_df.columns):
+            self.log.important(f"Target Column {primary_target} not found in prediction_df!")
             self.log.important("In order to compute metrics, the target column must be present!")
-            return prediction_df
+            metrics = pd.DataFrame()
 
         # Compute the standard performance metrics for this model
-        model_type = model.model_type
-        if model_type in [ModelType.REGRESSOR, ModelType.UQ_REGRESSOR, ModelType.ENSEMBLE_REGRESSOR]:
-            prediction_df = self.residuals(target_column, prediction_df)
-            metrics = self.regression_metrics(target_column, prediction_df)
-        elif model_type == ModelType.CLASSIFIER:
-            metrics = self.classification_metrics(target_column, prediction_df)
         else:
-            # For other model types, we don't compute metrics
-            self.log.info(f"Model Type: {model_type} doesn't have metrics...")
-            metrics = pd.DataFrame()
+            if model.model_type in [ModelType.REGRESSOR, ModelType.UQ_REGRESSOR, ModelType.ENSEMBLE_REGRESSOR]:
+                prediction_df = self.residuals(primary_target, prediction_df)
+                metrics = self.regression_metrics(primary_target, prediction_df)
+            elif model.model_type == ModelType.CLASSIFIER:
+                metrics = self.classification_metrics(primary_target, prediction_df)
+            else:
+                # For other model types, we don't compute metrics
+                self.log.info(f"Model Type: {model.model_type} doesn't have metrics...")
+                metrics = pd.DataFrame()
 
         # Print out the metrics
-        if not metrics.empty:
-            print(f"Performance Metrics for {self.model_name} on {self.name}")
-            print(metrics.head())
-
-            # Capture the inference results and metrics
-            if capture_name is not None:
-
-                # If we don't have an id_column, we'll pull it from the model's FeatureSet
-                if id_column is None:
-                    fs = FeatureSetCore(model.get_input())
-                    id_column = fs.id_column
-                description = capture_name.replace("_", " ").title()
+        print(f"Performance Metrics for {self.model_name} on {self.name}")
+        print(metrics.head())
+
+        # Capture the inference results and metrics
+        if capture_name is not None:
+
+            # If we don't have an id_column, we'll pull it from the model's FeatureSet
+            if id_column is None:
+                fs = FeatureSetCore(model.get_input())
+                id_column = fs.id_column
+
+            # Normalize targets to a list for iteration
+            target_list = targets if isinstance(targets, list) else [targets]
+
+            # For multi-target models, use target-specific capture names (e.g., auto_target1, auto_target2)
+            # For single-target models, use the original capture name for backward compatibility
+            for target in target_list:
+                # Determine capture name: use prefix for multi-target, original name for single-target
+                if len(target_list) > 1:
+                    prefix = "auto" if "auto" in capture_name else capture_name
+                    target_capture_name = f"{prefix}_{target}"
+                else:
+                    target_capture_name = capture_name
+
+                description = target_capture_name.replace("_", " ").title()
+
+                # Drop rows with NaN target values for metrics/plots
+                target_df = prediction_df.dropna(subset=[target])
+
+                # Compute per-target metrics
+                if model.model_type in [ModelType.REGRESSOR, ModelType.UQ_REGRESSOR, ModelType.ENSEMBLE_REGRESSOR]:
+                    target_metrics = self.regression_metrics(target, target_df)
+                elif model.model_type == ModelType.CLASSIFIER:
+                    target_metrics = self.classification_metrics(target, target_df)
+                else:
+                    target_metrics = pd.DataFrame()
+
                 self._capture_inference_results(
-                    capture_name, prediction_df, target_column, model_type, metrics, description, features, id_column
+                    target_capture_name,
+                    target_df,
+                    target,
+                    model.model_type,
+                    target_metrics,
+                    description,
+                    features,
+                    id_column,
                 )
 
-                # For UQ Models we also capture the uncertainty metrics
-                if model_type in [ModelType.UQ_REGRESSOR]:
-                    metrics = uq_metrics(prediction_df, target_column)
-                    self.param_store.upsert(f"/workbench/models/{model.name}/inference/{capture_name}", metrics)
+            # For UQ Models we also capture the uncertainty metrics
+            if model.model_type in [ModelType.UQ_REGRESSOR]:
+                metrics = uq_metrics(prediction_df, primary_target)
+                self.param_store.upsert(f"/workbench/models/{model.name}/inference/{capture_name}", metrics)
 
         # Return the prediction DataFrame
         return prediction_df
@@ -452,7 +493,16 @@ class EndpointCore(Artifact):
         model = ModelCore(self.model_name)
 
         # Compute CrossFold (Metrics and Prediction Dataframe)
-        cross_fold_metrics, out_of_fold_df = cross_fold_inference(model, nfolds=nfolds)
+        # For PyTorch and ChemProp, pull pre-computed CV results from training
+        if model.model_framework in [ModelFramework.UNKNOWN, ModelFramework.XGBOOST]:
+            cross_fold_metrics, out_of_fold_df = xgboost_cross_fold(model, nfolds=nfolds)
+        elif model.model_framework == ModelFramework.PYTORCH_TABULAR:
+            cross_fold_metrics, out_of_fold_df = pytorch_pull_cv(model)
+        elif model.model_framework == ModelFramework.CHEMPROP:
+            cross_fold_metrics, out_of_fold_df = chemprop_pull_cv(model)
+        else:
+            self.log.error(f"Cross-Fold Inference not supported for Model Framework: {model.model_framework}.")
+            return pd.DataFrame()
 
         # If the metrics dataframe isn't empty save to the param store
         if not cross_fold_metrics.empty:
@@ -460,10 +510,13 @@ class EndpointCore(Artifact):
             metrics = cross_fold_metrics.to_dict(orient="records")
             self.param_store.upsert(f"/workbench/models/{model.name}/inference/cross_fold", metrics)
 
+        # If the out_of_fold_df is empty return it
+        if out_of_fold_df.empty:
+            self.log.warning("No out-of-fold predictions were made. Returning empty DataFrame.")
+            return out_of_fold_df
+
         # Capture the results
-        capture_name = "full_cross_fold"
-        description = capture_name.replace("_", " ").title()
-        target_column = model.target()
+        targets = model.target()  # Note: We have multi-target models (so this could be a list)
         model_type = model.model_type
 
         # Get the id_column from the model's FeatureSet
@@ -472,7 +525,7 @@ class EndpointCore(Artifact):
 
         # Is this a UQ Model? If so, run full inference and merge the results
         additional_columns = []
-        if model_type == ModelType.UQ_REGRESSOR:
+        if model.model_framework == ModelFramework.XGBOOST and model_type == ModelType.UQ_REGRESSOR:
             self.log.important("UQ Regressor detected, running full inference to get uncertainty estimates...")
 
             # Get the training view dataframe for inference
@@ -481,9 +534,11 @@ class EndpointCore(Artifact):
             # Run inference on the endpoint to get UQ outputs
             uq_df = self.inference(training_df)
 
-            # Identify UQ-specific columns (quantiles and prediction_std)
+            # Identify UQ-specific columns (quantiles, prediction_std, *_pred_std)
             uq_columns = [
-                col for col in uq_df.columns if col.startswith("q_") or col == "prediction_std" or col == "confidence"
+                col
+                for col in uq_df.columns
+                if col.startswith("q_") or col == "prediction_std" or col.endswith("_pred_std") or col == "confidence"
             ]
 
             # Merge UQ columns with out-of-fold predictions
@@ -499,20 +554,42 @@ class EndpointCore(Artifact):
                 additional_columns = uq_columns
                 self.log.info(f"Added UQ columns: {', '.join(additional_columns)}")
 
-                # Also compute UQ metrics
-                metrics = uq_metrics(out_of_fold_df, target_column)
-                self.param_store.upsert(f"/workbench/models/{model.name}/inference/{capture_name}", metrics)
+                # Also compute UQ metrics (use first target for multi-target models)
+                primary_target = targets[0] if isinstance(targets, list) else targets
+                metrics = uq_metrics(out_of_fold_df, primary_target)
+                self.param_store.upsert(f"/workbench/models/{model.name}/inference/full_cross_fold", metrics)
+
+        # Normalize targets to a list for iteration
+        target_list = targets if isinstance(targets, list) else [targets]
+
+        # For multi-target models, use target-specific capture names (e.g., cv_target1, cv_target2)
+        # For single-target models, use "full_cross_fold" for backward compatibility
+        for target in target_list:
+            capture_name = f"cv_{target}"
+            description = capture_name.replace("_", " ").title()
+
+            # Drop rows with NaN target values for metrics/plots
+            target_df = out_of_fold_df.dropna(subset=[target])
+
+            # Compute per-target metrics
+            if model_type in [ModelType.REGRESSOR, ModelType.UQ_REGRESSOR, ModelType.ENSEMBLE_REGRESSOR]:
+                target_metrics = self.regression_metrics(target, target_df)
+            elif model_type == ModelType.CLASSIFIER:
+                target_metrics = self.classification_metrics(target, target_df)
+            else:
+                target_metrics = pd.DataFrame()
+
+            self._capture_inference_results(
+                capture_name,
+                target_df,
+                target,
+                model_type,
+                target_metrics,
+                description,
+                features=additional_columns,
+                id_column=id_column,
+            )
 
-        self._capture_inference_results(
-            capture_name,
-            out_of_fold_df,
-            target_column,
-            model_type,
-            cross_fold_metrics,
-            description,
-            features=additional_columns,
-            id_column=id_column,
-        )
         return out_of_fold_df
 
     def fast_inference(self, eval_df: pd.DataFrame, threads: int = 4) -> pd.DataFrame:
@@ -718,23 +795,47 @@ class EndpointCore(Artifact):
         combined = row_hashes.values.tobytes()
         return hashlib.md5(combined).hexdigest()[:hash_length]
 
+    @staticmethod
+    def _find_prediction_column(df: pd.DataFrame, target_column: str) -> Optional[str]:
+        """Find the prediction column in a DataFrame.
+
+        Looks for 'prediction' column first, then '{target}_pred' pattern.
+
+        Args:
+            df: DataFrame to search
+            target_column: Name of the target column (used for {target}_pred pattern)
+
+        Returns:
+            Name of the prediction column, or None if not found
+        """
+        # Check for 'prediction' column first (legacy/standard format)
+        if "prediction" in df.columns:
+            return "prediction"
+
+        # Check for '{target}_pred' format (multi-target format)
+        target_pred_col = f"{target_column}_pred"
+        if target_pred_col in df.columns:
+            return target_pred_col
+
+        return None
+
     def _capture_inference_results(
         self,
         capture_name: str,
         pred_results_df: pd.DataFrame,
-        target_column: str,
+        target: str,
         model_type: ModelType,
         metrics: pd.DataFrame,
         description: str,
         features: list,
         id_column: str = None,
     ):
-        """Internal: Capture the inference results and metrics to S3
+        """Internal: Capture the inference results and metrics to S3 for a single target
 
         Args:
             capture_name (str): Name of the inference capture
             pred_results_df (pd.DataFrame): DataFrame with the prediction results
-            target_column (str): Name of the target column
+            target (str): Target column name
             model_type (ModelType): Type of the model (e.g. REGRESSOR, CLASSIFIER)
             metrics (pd.DataFrame): DataFrame with the performance metrics
             description (str): Description of the inference results
@@ -765,28 +866,12 @@ class EndpointCore(Artifact):
         self.log.info(f"Writing metrics to {inference_capture_path}/inference_metrics.csv")
         wr.s3.to_csv(metrics, f"{inference_capture_path}/inference_metrics.csv", index=False)
 
-        # Grab the target column, prediction column, any _proba columns, and the ID column (if present)
-        output_columns = [target_column]
-        output_columns += [col for col in pred_results_df.columns if "prediction" in col]
-
-        # Add any _proba columns to the output columns
-        output_columns += [col for col in pred_results_df.columns if col.endswith("_proba")]
-
-        # Add any Uncertainty Quantile columns to the output columns
-        output_columns += [col for col in pred_results_df.columns if col.startswith("q_") or col == "confidence"]
-
-        # Add the ID column
-        if id_column and id_column in pred_results_df.columns:
-            output_columns.insert(0, id_column)
-
-        # Write the predictions to our S3 Model Inference Folder
-        self.log.info(f"Writing predictions to {inference_capture_path}/inference_predictions.csv")
-        subset_df = pred_results_df[output_columns]
-        wr.s3.to_csv(subset_df, f"{inference_capture_path}/inference_predictions.csv", index=False)
+        # Save the inference predictions for this target
+        self._save_target_inference(inference_capture_path, pred_results_df, target, id_column)
 
         # CLASSIFIER: Write the confusion matrix to our S3 Model Inference Folder
         if model_type == ModelType.CLASSIFIER:
-            conf_mtx = self.generate_confusion_matrix(target_column, pred_results_df)
+            conf_mtx = self.generate_confusion_matrix(target, pred_results_df)
             self.log.info(f"Writing confusion matrix to {inference_capture_path}/inference_cm.csv")
             # Note: Unlike other dataframes here, we want to write the index (labels) to the CSV
             wr.s3.to_csv(conf_mtx, f"{inference_capture_path}/inference_cm.csv", index=True)
@@ -796,6 +881,57 @@ class EndpointCore(Artifact):
         model = ModelCore(self.model_name)
         model._load_inference_metrics(capture_name)
 
+    def _save_target_inference(
+        self,
+        inference_capture_path: str,
+        pred_results_df: pd.DataFrame,
+        target: str,
+        id_column: str = None,
+    ):
+        """Save inference results for a single target.
+
+        Args:
+            inference_capture_path (str): S3 path for inference capture
+            pred_results_df (pd.DataFrame): DataFrame with prediction results
+            target (str): Target column name
+            id_column (str, optional): Name of the ID column
+        """
+        # Start with ID column if present
+        output_columns = []
+        if id_column and id_column in pred_results_df.columns:
+            output_columns.append(id_column)
+
+        # Add target column if present
+        if target and target in pred_results_df.columns:
+            output_columns.append(target)
+
+        # Build the output DataFrame
+        output_df = pred_results_df[output_columns].copy() if output_columns else pd.DataFrame()
+
+        # For multi-task: map {target}_pred -> prediction, {target}_pred_std -> prediction_std
+        # For single-task: just grab prediction and prediction_std columns directly
+        pred_col = f"{target}_pred"
+        std_col = f"{target}_pred_std"
+        if pred_col in pred_results_df.columns:
+            # Multi-task columns exist
+            output_df["prediction"] = pred_results_df[pred_col]
+            if std_col in pred_results_df.columns:
+                output_df["prediction_std"] = pred_results_df[std_col]
+        else:
+            # Single-task: grab standard prediction columns
+            for col in ["prediction", "prediction_std"]:
+                if col in pred_results_df.columns:
+                    output_df[col] = pred_results_df[col]
+            # Also grab any _proba columns and UQ columns
+            for col in pred_results_df.columns:
+                if col.endswith("_proba") or col.startswith("q_") or col == "confidence":
+                    output_df[col] = pred_results_df[col]
+
+        # Write the predictions to S3
+        output_file = f"{inference_capture_path}/inference_predictions.csv"
+        self.log.info(f"Writing predictions to {output_file}")
+        wr.s3.to_csv(output_df, output_file, index=False)
+
     def regression_metrics(self, target_column: str, prediction_df: pd.DataFrame) -> pd.DataFrame:
         """Compute the performance metrics for this Endpoint
         Args:
@@ -810,10 +946,28 @@ class EndpointCore(Artifact):
             self.log.warning("No predictions were made. Returning empty DataFrame.")
             return pd.DataFrame()
 
+        # Find the prediction column: "prediction" or "{target}_pred"
+        prediction_col = self._find_prediction_column(prediction_df, target_column)
+        if prediction_col is None:
+            self.log.warning(f"No prediction column found for target '{target_column}'")
+            return pd.DataFrame()
+
+        # Check for NaN values in target or prediction columns
+        if prediction_df[target_column].isnull().any() or prediction_df[prediction_col].isnull().any():
+            # Compute the number of NaN values in each column
+            num_nan_target = prediction_df[target_column].isnull().sum()
+            num_nan_prediction = prediction_df[prediction_col].isnull().sum()
+            self.log.warning(
+                f"NaNs Found: {target_column} {num_nan_target} and {prediction_col}: {num_nan_prediction}."
+            )
+            self.log.warning(
+                "NaN values found in target or prediction columns. Dropping NaN rows for metric computation."
+            )
+            prediction_df = prediction_df.dropna(subset=[target_column, prediction_col])
+
         # Compute the metrics
         try:
             y_true = prediction_df[target_column]
-            prediction_col = "prediction" if "prediction" in prediction_df.columns else "predictions"
             y_pred = prediction_df[prediction_col]
 
             mae = mean_absolute_error(y_true, y_pred)
@@ -849,7 +1003,13 @@ class EndpointCore(Artifact):
 
         # Compute the residuals
         y_true = prediction_df[target_column]
-        prediction_col = "prediction" if "prediction" in prediction_df.columns else "predictions"
+
+        # Find the prediction column: "prediction" or "{target}_pred"
+        prediction_col = self._find_prediction_column(prediction_df, target_column)
+        if prediction_col is None:
+            self.log.warning(f"No prediction column found for target '{target_column}'. Cannot compute residuals.")
+            return prediction_df
+
         y_pred = prediction_df[prediction_col]
 
         # Check for classification scenario
@@ -891,6 +1051,19 @@ class EndpointCore(Artifact):
         Returns:
             pd.DataFrame: DataFrame with the performance metrics
         """
+        # Find the prediction column: "prediction" or "{target}_pred"
+        prediction_col = self._find_prediction_column(prediction_df, target_column)
+        if prediction_col is None:
+            self.log.warning(f"No prediction column found for target '{target_column}'")
+            return pd.DataFrame()
+
+        # Drop rows with NaN predictions (can't compute metrics on missing predictions)
+        nan_mask = prediction_df[prediction_col].isna()
+        if nan_mask.any():
+            n_nan = nan_mask.sum()
+            self.log.warning(f"Dropping {n_nan} rows with NaN predictions for metrics calculation")
+            prediction_df = prediction_df[~nan_mask].copy()
+
         # Get the class labels from the model
         class_labels = ModelCore(self.model_name).class_labels()
         if class_labels is None:
@@ -903,7 +1076,6 @@ class EndpointCore(Artifact):
             self.validate_proba_columns(prediction_df, class_labels)
 
         # Calculate precision, recall, f1, and support, handling zero division
-        prediction_col = "prediction" if "prediction" in prediction_df.columns else "predictions"
         scores = precision_recall_fscore_support(
             prediction_df[target_column],
             prediction_df[prediction_col],
@@ -954,9 +1126,20 @@ class EndpointCore(Artifact):
         Returns:
             pd.DataFrame: DataFrame with the confusion matrix
         """
+        # Find the prediction column: "prediction" or "{target}_pred"
+        prediction_col = self._find_prediction_column(prediction_df, target_column)
+        if prediction_col is None:
+            self.log.warning(f"No prediction column found for target '{target_column}'")
+            return pd.DataFrame()
+
+        # Drop rows with NaN predictions (can't include in confusion matrix)
+        nan_mask = prediction_df[prediction_col].isna()
+        if nan_mask.any():
+            n_nan = nan_mask.sum()
+            self.log.warning(f"Dropping {n_nan} rows with NaN predictions for confusion matrix")
+            prediction_df = prediction_df[~nan_mask].copy()
 
         y_true = prediction_df[target_column]
-        prediction_col = "prediction" if "prediction" in prediction_df.columns else "predictions"
         y_pred = prediction_df[prediction_col]
 
         # Get model class labels

workbench/core/artifacts/feature_set_core.py

@@ -16,8 +16,9 @@ from sagemaker.feature_store.feature_store import FeatureStore
 from workbench.core.artifacts.artifact import Artifact
 from workbench.core.artifacts.data_source_factory import DataSourceFactory
 from workbench.core.artifacts.athena_source import AthenaSource
+from workbench.utils.deprecated_utils import deprecated
 
-from typing import TYPE_CHECKING, Optional, List, Union
+from typing import TYPE_CHECKING, Optional, List, Dict, Union
 
 from workbench.utils.aws_utils import aws_throttle
 
@@ -509,6 +510,71 @@ class FeatureSetCore(Artifact):
         ].tolist()
         return hold_out_ids
 
+    def set_sample_weights(
+        self,
+        weight_dict: Dict[Union[str, int], float],
+        default_weight: float = 1.0,
+        exclude_zero_weights: bool = True,
+    ):
+        """Configure training view with sample weights for each ID.
+
+        Args:
+            weight_dict: Mapping of ID to sample weight
+                - weight > 1.0: oversample/emphasize
+                - weight = 1.0: normal (default)
+                - 0 < weight < 1.0: downweight/de-emphasize
+                - weight = 0.0: exclude from training
+            default_weight: Weight for IDs not in weight_dict (default: 1.0)
+            exclude_zero_weights: If True, filter out rows with sample_weight=0 (default: True)
+
+        Example:
+            weights = {
+                'compound_42': 3.0,  # oversample 3x
+                'compound_99': 0.1,  # noisy, downweight
+                'compound_123': 0.0, # exclude from training
+            }
+            model.set_sample_weights(weights)  # zeros automatically excluded
+            model.set_sample_weights(weights, exclude_zero_weights=False)  # keep zeros
+        """
+        from workbench.core.views import TrainingView
+
+        if not weight_dict:
+            self.log.important("Empty weight_dict, creating standard training view")
+            TrainingView.create(self, id_column=self.id_column)
+            return
+
+        self.log.important(f"Setting sample weights for {len(weight_dict)} IDs")
+
+        # Helper to format IDs for SQL
+        def format_id(id_val):
+            return repr(id_val)
+
+        # Build CASE statement for sample_weight
+        case_conditions = [
+            f"WHEN {self.id_column} = {format_id(id_val)} THEN {weight}" for id_val, weight in weight_dict.items()
+        ]
+        case_statement = "\n        ".join(case_conditions)
+
+        # Build inner query with sample weights
+        inner_sql = f"""SELECT
+            *,
+            CASE
+                {case_statement}
+                ELSE {default_weight}
+            END AS sample_weight
+        FROM {self.table}"""
+
+        # Optionally filter out zero weights
+        if exclude_zero_weights:
+            zero_count = sum(1 for weight in weight_dict.values() if weight == 0.0)
+            custom_sql = f"SELECT * FROM ({inner_sql}) WHERE sample_weight > 0"
+            self.log.important(f"Filtering out {zero_count} rows with sample_weight = 0")
+        else:
+            custom_sql = inner_sql
+
+        TrainingView.create_with_sql(self, sql_query=custom_sql, id_column=self.id_column)
+
+    @deprecated(version=0.9)
     def set_training_filter(self, filter_expression: Optional[str] = None):
         """Set a filter expression for the training view for this FeatureSet
 
@@ -528,6 +594,7 @@ class FeatureSetCore(Artifact):
             self, id_column=self.id_column, holdout_ids=holdout_ids, filter_expression=filter_expression
         )
 
+    @deprecated(version="0.9")
     def exclude_ids_from_training(self, ids: List[Union[str, int]], column_name: Optional[str] = None):
         """Exclude a list of IDs from the training view
 
@@ -551,6 +618,7 @@ class FeatureSetCore(Artifact):
         # Apply the filter
         self.set_training_filter(filter_expression)
 
+    @deprecated(version="0.9")
     def set_training_sampling(
         self,
         exclude_ids: Optional[List[Union[str, int]]] = None,