Merge pull request #248 from PolicyEngine/la-calibration

Calibrate local authority microdata to ONS experimental data

Author: nikhilwoodruff

.github/workflows/pull_request.yaml

@@ -51,10 +51,17 @@ jobs:
           HUGGING_FACE_TOKEN: ${{ secrets.HUGGING_FACE_TOKEN }}
       - name: Build datasets
         run: make data
-      - name: Save calibration log
+      - name: Save calibration log (constituencies)
         uses: actions/upload-artifact@v4
         with:
-          name: calibration_log.csv
-          path: calibration_log.csv
+          name: constituency_calibration_log.csv
+          path: constituency_calibration_log.csv
+      
+      - name: Save calibration log (local authorities)
+        uses: actions/upload-artifact@v4
+        with:
+          name: la_calibration_log.csv
+          path: la_calibration_log.csv
+
       - name: Run tests
         run: make test

.github/workflows/push.yaml

@@ -58,11 +58,17 @@ jobs:
           HUGGING_FACE_TOKEN: ${{ secrets.HUGGING_FACE_TOKEN }}
       - name: Build datasets
         run: make data
-      - name: Save calibration log
+      - name: Save calibration log (constituencies)
         uses: actions/upload-artifact@v4
         with:
-          name: calibration_log.csv
-          path: calibration_log.csv
+          name: constituency_calibration_log.csv
+          path: constituency_calibration_log.csv
+      
+      - name: Save calibration log (local authorities)
+        uses: actions/upload-artifact@v4
+        with:
+          name: la_calibration_log.csv
+          path: la_calibration_log.csv
       - name: Run tests
         run: make test
       - name: Upload data

changelog_entry.yaml

@@ -0,0 +1,4 @@
+- bump: minor
+  changes:
+    added:
+    - Housing cost calibration to LAs.

policyengine_uk_data/datasets/create_datasets.py

@@ -128,28 +128,33 @@ def main():
                 area_count=650,
                 weight_file="parliamentary_constituency_weights.h5",
                 excluded_training_targets=[],
-                log_csv="calibration_log.csv",
+                log_csv="constituency_calibration_log.csv",
                 verbose=True,  # Enable nested progress display
                 area_name="Constituency",
                 get_performance=get_performance,
                 nested_progress=nested_progress,  # Pass the nested progress manager
             )
 
             from policyengine_uk_data.datasets.local_areas.local_authorities.calibrate import (
+                get_performance as get_la_performance,
+            )
+            from policyengine_uk_data.datasets.local_areas.local_authorities.loss import (
                 create_local_authority_target_matrix,
             )
 
             # Run calibration with verbose progress
             frs_calibrated_las = calibrate_local_areas(
                 dataset=frs,
+                epochs=512,
                 matrix_fn=create_local_authority_target_matrix,
                 national_matrix_fn=create_national_target_matrix,
                 area_count=360,
                 weight_file="local_authority_weights.h5",
                 excluded_training_targets=[],
-                log_csv="calibration_log.csv",
+                log_csv="la_calibration_log.csv",
                 verbose=True,  # Enable nested progress display
                 area_name="Local Authority",
+                get_performance=get_la_performance,
                 nested_progress=nested_progress,  # Pass the nested progress manager
             )
 

policyengine_uk_data/datasets/imputations/salary_sacrifice.py

@@ -93,20 +93,6 @@ def save_salary_sacrifice_model():
         }
     )
 
-    n_participants = (
-        train_df["pension_contributions_via_salary_sacrifice"] > 0
-    ).sum()
-    print(f"Training salary sacrifice model on {len(train_df)} observations")
-    print(
-        f"  With SS contributions: {n_participants} "
-        f"({n_participants / len(train_df):.1%})"
-    )
-    mean_amount = train_df.loc[
-        train_df["pension_contributions_via_salary_sacrifice"] > 0,
-        "pension_contributions_via_salary_sacrifice",
-    ].mean()
-    print(f"  Mean SS amount (participants): £{mean_amount:,.0f}")
-
     # Train QRF model
     model = QRF()
     model.fit(train_df[PREDICTORS], train_df[IMPUTATIONS])
@@ -166,17 +152,10 @@ def impute_salary_sacrifice(
 
     # Get indicator for who was asked
     if "salary_sacrifice_asked" not in dataset.person.columns:
-        print(
-            "Warning: salary_sacrifice_asked not in dataset, "
-            "skipping imputation"
-        )
         return dataset
 
     ss_asked = dataset.person.salary_sacrifice_asked.values
 
-    # Identify imputation candidates: those not asked about SS
-    not_asked = ss_asked == 0
-
     # Create prediction DataFrame for all records
     pred_df = pd.DataFrame(
         {
@@ -208,17 +187,4 @@ def impute_salary_sacrifice(
     # Update dataset
     dataset.person["pension_contributions_via_salary_sacrifice"] = final_ss
 
-    # Report results (no targeting - just descriptive)
-    weights = sim.calculate("person_weight").values
-    is_employee = employment_income > 0
-    total_ss = (final_ss * weights).sum()
-    participation_rate = ((final_ss > 0) * weights * is_employee).sum() / (
-        weights * is_employee
-    ).sum()
-
-    print("Salary sacrifice imputation results (pre-calibration):")
-    print(f"  Total SS contributions: £{total_ss / 1e9:.1f}bn")
-    print(f"  Employee participation rate: {participation_rate:.1%}")
-    print("  (Final totals depend on subsequent weight calibration)")
-
     return dataset

policyengine_uk_data/datasets/imputations/student_loans.py

@@ -75,17 +75,4 @@ def impute_student_loan_plan(
     # Store as the plan type
     dataset.person["student_loan_plan"] = plan
 
-    # Report imputation results
-    weights = sim.calculate("person_weight").values
-    total_with_loan = (has_student_loan * weights).sum()
-    plan_1_count = (plan_1_mask * weights).sum()
-    plan_2_count = (plan_2_mask * weights).sum()
-    plan_5_count = (plan_5_mask * weights).sum()
-
-    print("Student loan plan imputation results:")
-    print(f"  Total with student loan: {total_with_loan / 1e6:.2f}m")
-    print(f"  Plan 1 (pre-2012): {plan_1_count / 1e6:.2f}m")
-    print(f"  Plan 2 (2012-2023): {plan_2_count / 1e6:.2f}m")
-    print(f"  Plan 5 (2023+): {plan_5_count / 1e6:.2f}m")
-
     return dataset

policyengine_uk_data/datasets/local_areas/constituencies/calibrate.py

@@ -11,7 +11,7 @@
 def calibrate(
     dataset: UKSingleYearDataset,
     excluded_training_targets=[],
-    log_csv="calibration_log.csv",
+    log_csv="constituency_calibration_log.csv",
     verbose: bool = False,
 ):
     return calibrate_local_areas(

policyengine_uk_data/datasets/local_areas/local_authorities/calibrate.py

@@ -11,7 +11,7 @@
 def calibrate(
     dataset: UKSingleYearDataset,
     excluded_training_targets=[],
-    log_csv="calibration_log.csv",
+    log_csv="la_calibration_log.csv",
     verbose: bool = False,
 ):
     return calibrate_local_areas(