sdc-redistribute

Redistribute values between geographies — direct area-weighted interpolation and parcel-weighted redistribution.

Part of the Social Data Commons toolkit. Extracted from sdc_core.redistribute; used by SDC pipelines to move measures from one geographic vintage/level onto another (e.g. 2010 tracts → 2020 block groups), producing _geo10/_geo20-suffixed measures.

Install

uv add sdc-redistribute   # or: pip install sdc-redistribute

Quickstart

Split a tract's count onto two equal-area block groups (geometries generated inline so the example is self-contained):

import tempfile, pathlib
import geopandas as gpd
import pandas as pd
from shapely.geometry import box
from sdc_redistribute import redistribute_direct

tmp = pathlib.Path(tempfile.mkdtemp())
gpd.GeoDataFrame({"geoid": ["T1"]}, geometry=[box(0, 0, 2, 2)], crs="EPSG:4326").to_file(tmp / "tract.geojson", driver="GeoJSON")
gpd.GeoDataFrame({"geoid": ["BG1", "BG2"]}, geometry=[box(0, 0, 1, 2), box(1, 0, 2, 2)], crs="EPSG:4326").to_file(tmp / "bg.geojson", driver="GeoJSON")

source_df = pd.DataFrame({"geoid": ["T1"], "year": [2020], "measure": ["pop"], "value": [100.0]})
out = redistribute_direct(source_df, source_geo=tmp / "tract.geojson",
                          target_geos={"block_group": tmp / "bg.geojson"}, count_cols=["pop"])
print(out[["geoid", "measure", "value"]].to_string(index=False))
#  geoid    measure  value
#    BG1 pop_direct   50.0
#    BG2 pop_direct   50.0

Public API

• redistribute_direct — area-proportional redistribution between two geographies.
• redistribute_parcels — parcel-centroid-weighted redistribution.
• run_redistribution — high-level wrapper driven by a pipeline.yaml config block.

Documentation

• Introduction
• Method comparison
• API reference