Merge pull request #13 from ihmeuw/feature/rmbarber/gbd2025-preprocess-raking

update preprocess functionality to run for GBD2025 (mostly updates fo…

Author: rmbarber

src/rra_population_model/postprocess/raking_factors/runner.py

@@ -31,6 +31,8 @@ def load_admin_populations(
         year, quarter = (int(s) for s in time_point.split("q"))
         if RAKING_VERSION == "gbd_2023":
             next_year = min(year + 1, 2024)
+        elif RAKING_VERSION == "gbd_2025":
+            next_year = min(year + 1, 2026)
         else:
             next_year = min(year + 1, 2100)
         weight = (int(quarter) - 1) / 4

src/rra_population_model/preprocess/raking_data/metadata.py

@@ -92,8 +92,8 @@ class SUPPLEMENT:
     ZERO_POPULATION = "zero_population"
 
 
-def load_supplmental_metadata() -> pd.DataFrame:
-    return pd.DataFrame(
+def load_supplmental_metadata(gbd_version: str) -> pd.DataFrame:
+    supplmental_metadata = pd.DataFrame(
         [
             # This is a manual mapping of locations not present in the GBD hierarchy to the
             # GBD region in which they reside. We grab the additional iso3 and location name
@@ -345,3 +345,20 @@ def load_supplmental_metadata() -> pd.DataFrame:
             "category",
         ],
     )
+
+    if gbd_version == "2025":
+        # added French data in GBD 2025
+        is_fra_admin0 = supplmental_metadata["location_id"].isin(
+            [
+                338,  # French Guiana
+                350,  # Guadaloupe
+                363,  # Martinique
+                364,  # Mayotte
+                387,  # Reunion
+            ]
+        )
+        supplmental_metadata = supplmental_metadata.loc[~is_fra_admin0].reset_index(
+            drop=True
+        )
+
+    return supplmental_metadata

src/rra_population_model/preprocess/raking_data/runner.py

@@ -26,11 +26,13 @@ def raking_data_main(
     hierarchies = utils.load_hierarchies(pm_data, gbd_version=gbd_version)
     populations = utils.load_ihme_populations(pm_data, gbd_version=gbd_version)
     shapes = utils.load_shapes(pm_data, gbd_version=gbd_version)
-    supplemental_metadata = load_supplmental_metadata()
+    supplemental_metadata = load_supplmental_metadata(gbd_version=gbd_version)
 
     print("Building WPP data...")
-    wpp_version = "2024" if out_version == "gbd_2023" else "2022"
-    wpp = utils.load_wpp_populations(pm_data, wpp_version=wpp_version)
+    wpp_version = "2022" if out_version in ["gbd_2021", "fhs_2021"] else "2024"
+    wpp = utils.load_wpp_populations(
+        pm_data, wpp_version=wpp_version, gbd_version=gbd_version
+    )
     # Add GBD location and region ids to the WPP data by mapping on iso3 codes
     wpp = utils.add_gbd_metadata_to_wpp(
         wpp=wpp,
@@ -86,7 +88,9 @@ def raking_data_main(
 @click.command()
 @clio.with_output_directory(pmc.MODEL_ROOT)
 @clio.with_choice(
-    "out_version", allow_all=False, choices=["gbd_2023", "gbd_2021", "fhs_2021"]
+    "out_version",
+    allow_all=False,
+    choices=["gbd_2025", "gbd_2023", "gbd_2021", "fhs_2021"],
 )
 def raking_data(
     output_dir: str,

src/rra_population_model/preprocess/raking_data/utils.py

@@ -18,7 +18,7 @@
 
 
 def load_wpp_populations(
-    pm_data: PopulationModelData, wpp_version: str
+    pm_data: PopulationModelData, wpp_version: str, gbd_version: str
 ) -> pd.DataFrame:
     if wpp_version in ["2022", "2024"]:
         wpp = pm_data.load_gbd_raking_input("population", f"wpp_{wpp_version}")
@@ -29,6 +29,18 @@ def load_wpp_populations(
             # GBD treats Kosovo as part of Serbia at the admin0 level
             "SRB": ["SRB", "XKX"],
         }
+        if gbd_version == "2025":
+            # Starting in 2025, GBD separates Metropolitan France and overseas departments (DROM-COMs)
+            merge_map.update(
+                {
+                    "FRA_97896": ["FRA"],  # Metropolitan France
+                    "FRA_338": ["GUF"],  # French Guiana
+                    "FRA_350": ["GLP"],  # Guadeloupe
+                    "FRA_363": ["MTQ"],  # Martinique
+                    "FRA_364": ["MYT"],  # Mayotte
+                    "FRA_387": ["REU"],  # Reunion Island
+                }
+            )
         for target, sources in merge_map.items():
             mask = wpp["iso3"].isin(sources)
             merged = (