Skip to content

README.md

Latest commit

 

History

History
31 lines (22 loc) · 1.2 KB

README.md

File metadata and controls

31 lines (22 loc) · 1.2 KB
Raw

Groups

Group assignments are dicts (id:group), which are used by group-aware and cost-aware sampling methods.

Admin groups (state, county, district, region, etc.):

cd groups
python generate_groups.py \
  --datasets usavars_pop \
  --gdf_paths ../0_data/admin_gdfs/usavars/gdf_counties_population_2015.geojson \
  --id_cols id \
  --group_cols combined_county_id \
  --group_type counties

Admin geodataframes are built with usavars_generate_gdfs.py (USAVars, requires Census county shapefiles) or togo_generate_gdfs.py (Togo). India SECC ships its admin levels in the shapefile already, so no processing is needed.

Image clusters (KMeans over feature vectors):

python image_clusters.py --dataset usavars_population --feature_path /path/to/features.pkl --num_clusters 8

--dataset must be one of: usavars_population, usavars_treecover, india_secc, togo_ph_h20.

NLCD land-cover clusters (US only, requires a 2016 NLCD raster):

python nlcd_groups.py --input_dir /path/to/naip/tifs --nlcd_path /path/to/nlcd.tif --output_dir ./nlcd_out