Round explanation values to 4 decimal places in ForecastOperator (#1160)

Author: codeloop

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

@@ -573,34 +573,44 @@ def _save_report(
         if self.spec.generate_explanations:
             try:
                 if not self.formatted_global_explanation.empty:
+                    # Round to 4 decimal places before writing
+                    global_expl_rounded = self.formatted_global_explanation.copy()
+                    global_expl_rounded = global_expl_rounded.apply(
+                        lambda col: np.round(col, 4) if np.issubdtype(col.dtype, np.number) else col
+                    )
                     if self.spec.generate_explanation_files:
                         write_data(
-                            data=self.formatted_global_explanation,
+                            data=global_expl_rounded,
                             filename=os.path.join(
                                 unique_output_dir, self.spec.global_explanation_filename
                             ),
                             format="csv",
                             storage_options=storage_options,
                             index=True,
                         )
-                    results.set_global_explanations(self.formatted_global_explanation)
+                    results.set_global_explanations(global_expl_rounded)
                 else:
                     logger.warning(
                         f"Attempted to generate global explanations for the {self.spec.global_explanation_filename} file, but an issue occured in formatting the explanations."
                     )
 
                 if not self.formatted_local_explanation.empty:
+                    # Round to 4 decimal places before writing
+                    local_expl_rounded = self.formatted_local_explanation.copy()
+                    local_expl_rounded = local_expl_rounded.apply(
+                        lambda col: np.round(col, 4) if np.issubdtype(col.dtype, np.number) else col
+                    )
                     if self.spec.generate_explanation_files:
                         write_data(
-                            data=self.formatted_local_explanation,
+                            data=local_expl_rounded,
                             filename=os.path.join(
                                 unique_output_dir, self.spec.local_explanation_filename
                             ),
                             format="csv",
                             storage_options=storage_options,
                             index=True,
                         )
-                    results.set_local_explanations(self.formatted_local_explanation)
+                    results.set_local_explanations(local_expl_rounded)
                 else:
                     logger.warning(
                         f"Attempted to generate local explanations for the {self.spec.local_explanation_filename} file, but an issue occured in formatting the explanations."

tests/operators/forecast/test_explainers.py

@@ -343,8 +343,19 @@ def test_explanations_values(model, num_series, freq):
         if model == "automlx":
             pytest.xfail("automlx model does not provide fitted values")
 
+        # Check decimal precision for local explanations
+        local_numeric = local_explanations.select_dtypes(include=["int64", "float64"])
+        assert np.allclose(local_numeric, np.round(local_numeric, 4), atol=1e-8), \
+            "Local explanations have values with more than 4 decimal places"
+
+        # Check decimal precision for global explanations
+        global_explanations = results.get_global_explanations()
+        global_numeric = global_explanations.select_dtypes(include=["int64", "float64"])
+        assert np.allclose(global_numeric, np.round(global_numeric, 4), atol=1e-8), \
+            "Global explanations have values with more than 4 decimal places"
+
         local_explain_vals = (
-            local_explanations.select_dtypes(include=["int64", "float64"]).sum(axis=1)
+            local_numeric.sum(axis=1)
             + forecast.fitted_value.mean()
         )
         assert np.allclose(