fix: align missing service-line fallback

CHANGELOG.md

@@ -5,6 +5,12 @@ Format: [Keep a Changelog](https://keepachangelog.com/en/1.1.0/)
 
 ## [Unreleased]
 
+### Fixed
+- `GratefulPatientFeaturizer` now reports one fallback `general` service line
+  per known donor when the encounter table omits the service-line column.
+  Missing physician columns continue to report zero distinct physicians, and
+  both optional-column paths now have regression coverage. Closes #151.
+
 ### Changed
 - `WealthPercentileTransformer.fit` now raises an actionable `ValueError` when
   an explicit `wealth_cols` list matches no training column; partial matches

CONTRIBUTORS.md

@@ -56,6 +56,10 @@ contribution. Code, docs, tests, and review all count.
   schema mismatches fail with actionable diagnostics and preserved automatic
   and partial-match behavior (closes
   [#156](https://github.com/PhilanthroPy-Project/PhilanthroPy/issues/156)).
+- [@be-student](https://github.com/be-student): aligned the missing service-line
+  fallback with its `general` category and added regression coverage for
+  missing service-line and physician columns (closes
+  [#151](https://github.com/PhilanthroPy-Project/PhilanthroPy/issues/151)).
 
 ## Getting listed
 

philanthropy/preprocessing/_grateful_patient.py

@@ -278,7 +278,7 @@ def fit(self: _Self, X: Any, y: Any = None) -> _Self:
                 "general", index=grouped.groups.keys()
             )
             summary_parts["distinct_service_lines"] = pd.Series(
-                0, dtype=int, index=grouped.groups.keys()
+                1, dtype=int, index=grouped.groups.keys()
             )
 
         if self.physician_col in raw_enc.columns:

tests/test_grateful_patient_featurizer.py

@@ -133,6 +133,32 @@ def test_missing_drg_weight_col_fills_nan_then_zero_after_transform(
             "Without drg_weight_col, total_drg_weight must be 0.0 after fillna"
         )
 
+    def test_missing_service_line_col_falls_back_to_general(
+        self, enc_df, X_donors
+    ):
+        enc = enc_df.drop(columns=["service_line"])
+        gpf = GratefulPatientFeaturizer(encounter_df=enc).fit(X_donors)
+
+        result = gpf.transform(X_donors)
+        feature_names = list(gpf.get_feature_names_out())
+        distinct_idx = feature_names.index("distinct_service_lines")
+
+        assert result.shape == (len(X_donors), len(feature_names))
+        assert np.isfinite(result).all()
+        np.testing.assert_array_equal(result[:4, distinct_idx], 1.0)
+
+    def test_missing_physician_col_yields_zero_distinct_physicians(
+        self, enc_df, X_donors
+    ):
+        enc = enc_df.drop(columns=["attending_physician_id"])
+        gpf = GratefulPatientFeaturizer(encounter_df=enc).fit(X_donors)
+
+        result = gpf.transform(X_donors)
+        feature_names = list(gpf.get_feature_names_out())
+        physician_idx = feature_names.index("distinct_physicians")
+
+        np.testing.assert_array_equal(result[:4, physician_idx], 0.0)
+
     def test_drg_weight_col_sums_correctly(self, X_donors):
         """With drg_weight_col present, total_drg_weight must sum DRG weights per donor."""
         enc = pd.DataFrame({