ABSTRACT
Chapter 8 covers the use of Census data to derive models for explaining patterns that occur across regions or within cities. The first section looks at segregation and diversity indices which are widely used across the social sciences to explain demographic patterns. The second section explores topics in statistical modeling, including methods for spatial regression that take into account the spatial autocorrelation inherent in most Census variables. The third and final section explores concepts such as classification, clustering, and regionalization which are common in both unsupervised and supervised machine learning. Examples will illustrate how to use Census data to generate neighborhood typologies, which are widely used for business and marketing applications, and how to generate spatially coherent sales territories from Census data with regionalization.
