Modeling Gentrification

I am creating a model of gentrification for my senior capstone project. I plan to create machine learning (ML) model that I will apply some neighborhoods in King County, to inform an agent-based model that can be more widely applied and potentially evaluate the impact of different interventions for community resilience.

Motivation

For a long time I have been curious about our relationship with space and how we live on stolen land, particularly how people migrate within the US and what catalyzes displacement and exodus. Living in Seattle I have witnessed a sudden dynamic shift in neighbors and businesses in the city, and how that has impacted marginalized communities. I want to create a model that can identify communities vulnerable to gentrification, prior to the violence of displacement. By providing this sort of signaling mechanism I think it will reinforce the self-determination and resilience of neighborhoods. Though I don't think this analysis will reveal anything that we haven't heard on the stoop or from critical scholars; I do believe it will be a substantial tool in illustrating conditions and supporting organizing efforts at the grassroots level to catalyze change.

Intention

I intend to develop ML models to identify some determinant factors that make communities vulnerable to gentrification. Then I would like to develop an agent-based model to understand the dynamics more deeply and provide a general model to articulate gentrification and potentially evaluate the impact of different interventions.

Process

Currently I'm working on running the code Reades, De Souza, and Hubbard shared in, "Understanding urban gentrification through machine learning".

Data Being Used

I'm beginning my analysis with Census data at the block group level, this is helpful as it is a nationally implemented method of collecting data on people living in the US. However, the history of the Census, mistrust of how the government may use data collected, and implementation results in marginalized populations being underrepresented in the data. Across the decades there are also changes to the questionnaire and the boundaries of block groups. I would like to primarily rely on data sets that are generally available and collected across the country so this project is replicable in other areas, and can capture communities underrepresented by the Census.

Historical Grounding

The Pacific Northwest has strong history of racial exclusion and violence that extends beyond the Manifest Destiny employed by early settlers. In terms of folks accessing housing, the racial exclusion acts, practice of redlining, and racial restrictive covenants reinforced residential segregation and perpetuated disparities in accessing affordable housing and land. The burden of this discrimination continues to manifest today in the quality of life in neighborhoods as well as how resources are allocated throughout the city. After years of marginalization communities of color are being pushed out of their neighborhoods through gentrification and urban renewal.

Home Owners' Loan Corporation (HOLC) map of the Greater Seattle area that details the redlining

Descriptions of select areas in the HOLC map

A map from the Communist paper, The New World, highlighting the segregation BIPOC communities were facing

Where and when?

I am particularly interested in modeling Black communities in the Greater Seattle area so I will be investigating dynamics in neighborhoods such as the Central District, Yesler Terrace, and South End of Seattle to train and test my ML model. In order to capture more representative data I will begin my analysis at the close of the Great Migration. Beginning here captures the experience of Black folks in accessing housing and land after the repeal of Black exclusion laws and passage of the Housing Act.

Where do neighborhoods begin and end?

Neighborhood borders are fluid and not consistently defined across institutions or over time. The boundaries of neighborhoods shift for a variety of reasons, and the way we choose to draw these maps are powerful in that they imply claim to space and spheres of influence. In exploring how the Central District is defined I came across a number of maps, each of which has bias or incentive to draw the neighborhood a particular way. Especially in the context of gentrification when many of the maps we are often exposed to are drawn by privately owned companies (e.g. Google, Zillow, Airbnb) with interests in how these communities are defined. Below are a few maps of the Central district from recent years from different sources.

Seattle Strategic Planning Office 1999

Seattle City Clerk 2018

Google Maps 2020

Race Heat Map of Population in the Greater Seattle Area

These maps details the percentage of Black and white residents of the Greater Seattle area by Census tract from 1970-2010. Maps are sourced from the amazing work by the Seattle Civil Right & Labor History Project at the University of Washington.