From Individual Tracts to Community Segments: an Unsupervised Learning Approach for the Chicago Community Areas

This paper presents the results of a cross-sectional study that analyzes housing characteristics and the demographic makeup of Chicago community areas. The study is part of a larger project that uses data mining techniques to evaluate and model changes in neighborhoods that may lead to gentrification or abandonment, and ultimately to the loss of affordable housing. Our initial analysis has defined a typology for the seventy-seven Chicago community areas by classifying them in five segments according to median income level, education, race, and crime rates. Such a classification is important to understand similarities among the communities and to predict where changes are more likely to occur. For instance it can be hypothesized that gentrification is in progress for those communities that do not fit in just one type, but show subareas that belong to a higher-income segment.