We have developed an algorithm that reduces the margins of error in ACS Tract and Block Group Data to a user specified level by “intelligently” combining Census geographies together into regions. A region is a collection of 1 or more census geographies that meets a user specified margin of error (or CV). We refer to this procedure as “regionalization."
Tutorials: We have developed a tutorial using an extremely simple toy example (Toy Example) and a more realistic tutorial using data for Austin, TX (Austin Example).