Merge branch 'main' into dataflow_changes

Author: ahosler

ads/opctl/operator/lowcode/forecast/const.py

@@ -27,10 +27,12 @@ class SpeedAccuracyMode(str, metaclass=ExtendedEnumMeta):
     HIGH_ACCURACY = "HIGH_ACCURACY"
     BALANCED = "BALANCED"
     FAST_APPROXIMATE = "FAST_APPROXIMATE"
+    AUTOMLX = "AUTOMLX"
     ratio = {}
     ratio[HIGH_ACCURACY] = 1  # 100 % data used for generating explanations
     ratio[BALANCED] = 0.5  # 50 % data used for generating explanations
     ratio[FAST_APPROXIMATE] = 0  # constant
+    ratio[AUTOMLX] = 0  # constant
 
 
 class SupportedMetrics(str, metaclass=ExtendedEnumMeta):

ads/opctl/operator/lowcode/forecast/model/automlx.py

@@ -17,6 +17,7 @@
 from ads.opctl.operator.lowcode.forecast.const import (
     AUTOMLX_METRIC_MAP,
     ForecastOutputColumns,
+    SpeedAccuracyMode,
     SupportedModels,
 )
 from ads.opctl.operator.lowcode.forecast.utils import _label_encode_dataframe
@@ -245,18 +246,18 @@ def _generate_report(self):
                 # If the key is present, call the "explain_model" method
                 self.explain_model()
 
-                # Convert the global explanation data to a DataFrame
-                global_explanation_df = pd.DataFrame(self.global_explanation)
+                global_explanation_section = None
+                if self.spec.explanations_accuracy_mode != SpeedAccuracyMode.AUTOMLX:
+                    # Convert the global explanation data to a DataFrame
+                    global_explanation_df = pd.DataFrame(self.global_explanation)
 
-                self.formatted_global_explanation = (
-                    global_explanation_df / global_explanation_df.sum(axis=0) * 100
-                )
-                self.formatted_global_explanation = (
-                    self.formatted_global_explanation.rename(
+                    self.formatted_global_explanation = (
+                        global_explanation_df / global_explanation_df.sum(axis=0) * 100
+                    )
+                    self.formatted_global_explanation = self.formatted_global_explanation.rename(
                         {self.spec.datetime_column.name: ForecastOutputColumns.DATE},
                         axis=1,
                     )
-                )
 
                 aggregate_local_explanations = pd.DataFrame()
                 for s_id, local_ex_df in self.local_explanation.items():
@@ -297,8 +298,11 @@ def _generate_report(self):
                 )
 
                 # Append the global explanation text and section to the "other_sections" list
+                if global_explanation_section:
+                    other_sections.append(global_explanation_section)
+
+                # Append the local explanation text and section to the "other_sections" list
                 other_sections = other_sections + [
-                    global_explanation_section,
                     local_explanation_section,
                 ]
             except Exception as e:
@@ -379,3 +383,79 @@ def _custom_predict_automlx(self, data):
         return self.models.get(self.series_id).forecast(
             X=data_temp, periods=data_temp.shape[0]
         )[self.series_id]
+
+    @runtime_dependency(
+        module="automlx",
+        err_msg=(
+            "Please run `python3 -m pip install automlx` to install the required dependencies for model explanation."
+        ),
+    )
+    def explain_model(self):
+        """
+        Generates explanations for the model using the AutoMLx library.
+
+        Parameters
+        ----------
+        None
+
+        Returns
+        -------
+        None
+
+        Notes
+        -----
+        This function works by generating local explanations for each series in the dataset.
+        It uses the ``MLExplainer`` class from the AutoMLx library to generate feature attributions
+        for each series. The feature attributions are then stored in the ``self.local_explanation`` dictionary.
+
+        If the accuracy mode is set to AutoMLX, it uses the AutoMLx library to generate explanations.
+        Otherwise, it falls back to the default explanation generation method.
+        """
+        import automlx
+
+        # Loop through each series in the dataset
+        for s_id, data_i in self.datasets.get_data_by_series(
+            include_horizon=False
+        ).items():
+            try:
+                if self.spec.explanations_accuracy_mode == SpeedAccuracyMode.AUTOMLX:
+                    # Use the MLExplainer class from AutoMLx to generate explanations
+                    explainer = automlx.MLExplainer(
+                        self.models[s_id],
+                        self.datasets.additional_data.get_data_for_series(series_id=s_id)
+                        .drop(self.spec.datetime_column.name, axis=1)
+                        .head(-self.spec.horizon)
+                        if self.spec.additional_data
+                        else None,
+                        pd.DataFrame(data_i[self.spec.target_column]),
+                        task="forecasting",
+                    )
+
+                    # Generate explanations for the forecast
+                    explanations = explainer.explain_prediction(
+                        X=self.datasets.additional_data.get_data_for_series(series_id=s_id)
+                        .drop(self.spec.datetime_column.name, axis=1)
+                        .tail(self.spec.horizon)
+                        if self.spec.additional_data
+                        else None,
+                        forecast_timepoints=list(range(self.spec.horizon + 1)),
+                    )
+
+                    # Convert the explanations to a DataFrame
+                    explanations_df = pd.concat(
+                        [exp.to_dataframe() for exp in explanations]
+                    )
+                    explanations_df["row"] = explanations_df.groupby("Feature").cumcount()
+                    explanations_df = explanations_df.pivot(
+                        index="row", columns="Feature", values="Attribution"
+                    )
+                    explanations_df = explanations_df.reset_index(drop=True)
+
+                    # Store the explanations in the local_explanation dictionary
+                    self.local_explanation[s_id] = explanations_df
+                else:
+                    # Fall back to the default explanation generation method
+                    super().explain_model()
+            except Exception as e:
+                logger.warning(f"Failed to generate explanations for series {s_id} with error: {e}.")
+                logger.debug(f"Full Traceback: {traceback.format_exc()}")

ads/opctl/operator/lowcode/forecast/model/base_model.py

@@ -49,6 +49,7 @@
     SpeedAccuracyMode,
     SupportedMetrics,
     SupportedModels,
+    BACKTEST_REPORT_NAME,
 )
 from ..operator_config import ForecastOperatorConfig, ForecastOperatorSpec
 from .forecast_datasets import ForecastDatasets
@@ -665,6 +666,13 @@ def _save_model(self, output_dir, storage_options):
             storage_options=storage_options,
         )
 
+    def _validate_automlx_explanation_mode(self):
+        if self.spec.model != SupportedModels.AutoMLX and self.spec.explanations_accuracy_mode == SpeedAccuracyMode.AUTOMLX:
+            raise ValueError(
+                "AUTOMLX explanation accuracy mode is only supported for AutoMLX models. "
+                "Please select mode other than AUTOMLX from the available explanations_accuracy_mode options"
+            )
+
     @runtime_dependency(
         module="shap",
         err_msg=(
@@ -693,6 +701,9 @@ def explain_model(self):
         )
         ratio = SpeedAccuracyMode.ratio[self.spec.explanations_accuracy_mode]
 
+        # validate the automlx mode is use for automlx model
+        self._validate_automlx_explanation_mode()
+
         for s_id, data_i in self.datasets.get_data_by_series(
             include_horizon=False
         ).items():
@@ -727,6 +738,14 @@ def explain_model(self):
                     logger.warn(
                         "No explanations generated. Ensure that additional data has been provided."
                     )
+                elif (
+                    self.spec.model == SupportedModels.AutoMLX
+                    and self.spec.explanations_accuracy_mode
+                    == SpeedAccuracyMode.AUTOMLX
+                ):
+                    logger.warning(
+                        "Global explanations not available for AutoMLX models with inherent explainability"
+                    )
                 else:
                     self.global_explanation[s_id] = dict(
                         zip(

ads/opctl/operator/lowcode/forecast/schema.yaml

@@ -344,6 +344,7 @@ spec:
         - HIGH_ACCURACY
         - BALANCED
         - FAST_APPROXIMATE
+        - AUTOMLX
 
     generate_report:
       type: boolean

tests/operators/forecast/test_errors.py

@@ -591,6 +591,8 @@ def test_all_series_failure(model):
     yaml_i["spec"]["preprocessing"] = {"enabled": True, "steps": preprocessing_steps}
     if yaml_i["spec"].get("additional_data") is not None and model != "autots":
         yaml_i["spec"]["generate_explanations"] = True
+    else:
+        yaml_i["spec"]["generate_explanations"] = False
     if model == "autots":
         yaml_i["spec"]["model_kwargs"] = {"model_list": "superfast"}
     if model == "automlx":
@@ -672,6 +674,7 @@ def test_arima_automlx_errors(operator_setup, model):
         yaml_i["spec"]["model_kwargs"] = {"model_list": "superfast"}
     if model == "automlx":
         yaml_i["spec"]["model_kwargs"] = {"time_budget": 1}
+        yaml_i["spec"]["explanations_accuracy_mode"] = "AUTOMLX"
 
     run_yaml(
         tmpdirname=tmpdirname,
@@ -699,21 +702,15 @@ def test_arima_automlx_errors(operator_setup, model):
                 in error_content["13"]["error"]
             ), "Error message mismatch"
 
-    if model not in ["autots", "automlx"]:  # , "lgbforecast"
-        global_fn = f"{tmpdirname}/results/global_explanation.csv"
-        assert os.path.exists(
-            global_fn
-        ), f"Global explanation file not found at {report_path}"
+    if model not in ["autots"]:  # , "lgbforecast"
+        if yaml_i["spec"].get("explanations_accuracy_mode") != "AUTOMLX":
+            global_fn = f"{tmpdirname}/results/global_explanation.csv"
+            assert os.path.exists(global_fn), f"Global explanation file not found at {report_path}"
+            assert not pd.read_csv(global_fn, index_col=0).empty
 
         local_fn = f"{tmpdirname}/results/local_explanation.csv"
-        assert os.path.exists(
-            local_fn
-        ), f"Local explanation file not found at {report_path}"
-
-        glb_expl = pd.read_csv(global_fn, index_col=0)
-        loc_expl = pd.read_csv(local_fn)
-        assert not glb_expl.empty
-        assert not loc_expl.empty
+        assert os.path.exists(local_fn), f"Local explanation file not found at {report_path}"
+        assert not pd.read_csv(local_fn).empty
 
 
 def test_smape_error():