An open-source framework for non-Spatial and Spatial segregation measures: the PySAL segregation module

Renan X. Cortes, Sergio Rey, Elijah Knaap, Levi John Wolf

In urban geography and social sciences, segregation, usually consider five dimensions ina given society such as evenness, isolation, clustering, concentration and centralization. Allof these measure can either ignore spatial context or take it into consideration. Currently,several segregation measures are available in the literature, but they lack of wide spreaduse, in part, due to their complex calculations. In addition, there are only a few works thataddress the problem of inference in segregation measures for either single measure or forcomparison between multiple measures. This work tries to fill this gap by constructing anopen-source segregation module in the Python Spatial Analysis Library (PySAL). This newmodule tackles the problem of segregation point estimation for some well-known non-spatialsegregation indexes such as Dissimilarity (and its related), Gini, Entropy, Isolation, Concen-tration Profile, Correlation Ratio, and spatial indexes such as Spatial Proximity, RelativeClustering, Relative Concentration, Relative Centralization. Furthermore, it also presents anovel feature that performs inference for segregation and for comparative segregation, relyingon simulations under the null hypothesis. We illustrate the use of this new library using tractlevel census data in American counties of non-Hispanic black population.