Click the link above to access all EA Insights

By: Matt Mowell, Jamie Portolese, Franz Limoges

The U.S. Census publishes state and county-level data as well as for smaller areas like tracts, which contain blocks and block groups.  Analyzing data across tracts can be challenging because some city-center tracts are very small while those in the suburbs can be very large. 

We improve data analysis by converting tract and block group information into uniform hexagon shapes, known as Uber H3 hexagons. While Uber H3 hexagons are available in 16 different sizes, called resolutions, when selecting a resolution for an analysis, the hexagons at that resolution are always of a uniform size.

Here's the methodology: 

• We take the census areas (tracts and block groups) and find their geographic center (the middle point).
• We assign the census data, such as median household income, to these center points.
• We then use an interpolation technique called kriging to create a smooth surface layer of the data across the entire country. This helps us estimate income values between the points.
• Finally, we overlay hexagon shapes onto the surface to calculate average income (derived from an average of the surface layer created above) within each hexagon area.
• This lets us transfer information from census tracts or block groups to hexagons of different resolutions based on the granularity of the data required.

The hexagon shapes enable us to analyze census data in a more consistent way than we can when relying on tracts or block groups.