Using Machine Learning to Identify and Predict Gentrification in Nashville, Tennessee.
Knorr, David Christopher
:
2019-07-24
Abstract
Gentrification is a polarizing and elusive type of neighborhood change that disproportionately threatens our community’s most vulnerable populations. The lived consequences of gentrification have merited a substantial amount of academic focus over the course of more than five decades. This paper takes a critical examination of the prescriptive frameworks that previous researchers have used to distinguish gentrification, identifying errors of inclusion, exclusion, as well as major methodological inconsistencies. We advance a k-means clustering approach of six change variables to identify four dominant trajectories of neighborhood change in Nashville, TN (Davidson County) between 2000 and 2016. One specific typology indicated the tell-tale patterns of gentrification and evidence for residential displacement in 13% of inner-city census tracts. A significant join-count statistic revealed that these change typologies were clustered spatially; the observed spatial phenomena gives promise to the potential to predict gentrification as a rational process. We used these outcomes to train a random forest binary classification model to predict susceptibility to gentrification based on starting housing, demographic, transportation, amenity, and locational characteristics. The mapped predictions of gentrification reveal continuing gentrification in south and east Nashville as well as possible expansion beyond previously identified areas, predominantly along highway corridors to the north and southeast where the majority of the affordable housing stock remains. This research contributes to the quantitative efforts to identify gentrification by advancing a more holistic and less biased alternative. This work also contributes to the forward-looking literature on gentrification. It is designed to serve communities as well as policymakers to optimize intervention strategies in an effort to increase equitable development and socially just cities.