Add PUF datasets to outputs

Author: nikhilwoodruff

policyengine_us_data/datasets/puf/puf.py

@@ -357,11 +357,7 @@ def generate(self):
 
         i = 0
         self.earn_splits = []
-        for _, row in tqdm(
-            puf.iterrows(),
-            total=len(puf),
-            desc="Constructing hierarchical PUF",
-        ):
+        for _, row in puf.iterrows():
             i += 1
             exemptions = row["exemptions_count"]
             tax_unit_id = row["household_id"]
@@ -517,3 +513,5 @@ class PUF_2024(PUF):
 
 if __name__ == "__main__":
     PUF_2015().generate()
+    PUF_2021().generate()
+    PUF_2024().generate()

policyengine_us_data/utils/loss.py

@@ -236,7 +236,7 @@ def build_loss_matrix(dataset: type, time_period):
 
     loss_matrix["irs/negative_household_market_income_count"] = (
         market_income < 0
-    )
+    ).astype(float)
     targets_array.append(3e6)
 
     # Healthcare spending by age