Neighborhood Racial and Ethnic Trends in U.S. Cities: Mapping 30 Years of Change
Welcome to the Interactive Dashboard of Neighborhood !
Over the past several decades, the U.S. population has become increasingly racially and ethnically diverse.
These national demographic trends have reshaped local neighborhoods, but with considerable variation depending on metropolitan-scale population dynamics.
This interactive dashboard brings to life a comprehensive, data-driven study of how neighborhood racial and ethnic compositions have evolved in 66 U.S. metropolitan areas (MSA).
Using a time-series clustering approach with Dynamic Time Warping (DTW) and Partitioning Around Medoids (PAM),
the research behind aims to uncovers the complex pathways of demographic change at the neighborhood level—from the steady diversification of historically homogenous areas to the emergence of new majority-minority communities.
Navigate through the maps and trend charts to discover:
- National patterns through MSA scale and single clustering view
- Regional influence on local trends in our double clustering view
- City-specific stories using the MSA specific clustering view
- Neighborhood-specific stories through entering an address
Click on to explore the trends for national MSA Scale clustering, or click on a specific MSA polygon to dive into local trends.
This project is based on the paper “Time Series Clustering for Exploring Neighborhood Dynamics: The Case of U.S. Neighborhood Racial and Ethnic Trends, 1990–2020.” The code and methods are open-source to encourage further exploration and adaptation.
MSA Scale Clustering
The metropolitan statistical areas (MSA) have been classified into 6 groups based on similarities in racial and ethnic trends.
Time-series clustering is applied to each demographic group, and the clustering results are combined using a cross-classification approach.
All possible combinations of the clustering results are identified and assigned to each MSA, resulting in 6 main groups of MSAs, of which the top 3
are the main focus on analysis: predominantly White, White and Black, and Hispanic strong holds.
Click on each MSA polygon to learn more about the neighborhood-level demographics.
Single Clustering
Time-series clustering is applied to each demographic group on a national scale by census tract, and all unique combinations of clustering results are identified.
Most tracts (41%) had a very high share of Whites that experienced mild declines, accompanied by small increases in Blacks and Hispanics.
22% of tracts underwent more dramatic declines in Whites, falling to minority-White by 2020, with some accompanied by a sharp uptick in
Hispanics and some by more modest increases in Blacks.
14% of tracts have a high and increasing share of Hispanics or a high but declining share of Blacks.
Double Clustering
Time-series clustering is applied to each demographic group on a regional scale, by clustering on tracts that belong to the same MSA scale groups. A high but declining share of Whites remains the top trajectory, but with variations across MSA scale clusters. In predominantly White MSAs, there was a slight increase in Hispanics in the last decade, while in the White and Black MSAs, a high but declining share of Whites is accompanied by an uptick in Black residents. In the Hispanic MSAs, the share of Whites in 1990 is lower and touches the 50% mark by 2020. This is accompanied by an increasing share of Hispanics.
MSA-Specific Clustering
Time-series clustering is applied to each demographic group on a local scale by each MSA, with the addition of the Asian population. The MSA-specific clustering results show more complex details,
sometimes revealing tracts where other ethnicities like Asians are more dominant.
The clustering results are arranged from the most common to the least common in each MSA, both on the map and in the corresponding time series graphs.