The Neighborhood Change Indicators project used machine learning to predict future neighborhood change. The project built on the Urban Institute's previous work to expand the analysis from four metropolitan areas to nationwide. Machine learning algorithms were trained on HUD administrative data, USPS administrative data, and several other datasets to predict for specific, binary neighborhood outcomes. The neighborhood change outcome definitions were guided by a literature review and a Project Advisory Group of local experts and advocacy organizations on neighborhood change. Forecasting neighborhood change can help communities and policymakers understand which places may experience further change or are at risk of displacement or disinvestment. The results from this analysis will be available to help guide users of the Neighborhood Change Web Map. Predictions and actual results for 2022 can be viewed online.

Final Spatial Dataset Codebook

Final Neighborhood Change Indicators Presentation Slides