Tract Clustering Finding clusters of neighborhoods

Clustering Maryland Census Tracts

We downloaded joint income and age distribution from the 2009-2013 American Community Survey for all census tracts in Maryland. We compared these distributions by Hellinger distance, projected them into coordinate space using multidimensional scaling and then clustered them with k-medoids. Clusters appear to primarily reflect well known economic divides throughout Maryland.

Project Details

  • Census API with Python wrapper
  • Analysis in Matlab
  • TIGER Shapefiles via Python requests
  • Geodata cleaning and merging with Python geopandas
  • Map Visualization in Mapbox