Multivariate Neighborhood Trajectory Analysis: An Exploration of the Functional Data Analysis Approach

Recent neighborhood studies have focused on longitudinal aspects of neighborhood change and data-mining methodologies that identify neighborhood trajectory patterns using time-series multivariate census data. Existing neighborhood trajectory models capture neighborhood change by stacking cross-sectional neighborhood clustering results across years and analyzing the discrete stepwise switching patterns between the clusters. Taking a different approach, we employ the functional data analysis (FDA) method to analyze longitudinal patterns of neighborhood change from mathematically represented multivariate time-dependent curves to identify neighborhood trajectory clusters. This FDA-based neighborhood trajectory model incorporates a multivariate functional principal component analysis and k-means clustering. We have applied our model to neighborhoods in the Charlotte and Detroit metropolitan areas to identify ongoing racial and socioeconomic segregation patterns and the time dynamics of neighborhood change.

Automated summary

Recent neighborhood studies have focused on longitudinal aspects of neighborhood change and data-mining methodologies. Using functional data analysis method, neighborhood trajectory clusters are identified through mathematically represented multivariate time-dependent curves. The model has been applied to Charlotte and Detroit areas to analyze ongoing racial and socioeconomic segregation patterns and time dynamics of neighborhood change.