Classifying multidimensional trajectories of neighbourhood change: a self-organizing map andk-means approach

Abstract

Understanding the dynamic nature of how urban neighbourhoods evolve through time has been a critical issue both in the literature and in public policy practice for decades. Methodological limitations in understanding change across the multiple attribute dimensions that define a neighbourhood, through time and for spatially situated units, have largely reduced empirical analyses to two points in time or for a singular attribute dimension. This paper demonstrates a two-layered approach to classifying neighbourhoods according to their multidimensional, temporal trajectories. The method first projects data onto a two-dimensional output space using a self-organizing map, then constructs temporal trajectories of change across this space and, finally, classifies the resulting trajectories with a k-means algorithm. The resulting typology of neighbourhood trajectories are then mapped in the geographic space to visualize the space-time, multidimensional dynamics. A case study of neighbourhood change from 1970 to 2010 in eight US cities demonstrates the effectiveness approach for grouping neighbourhoods according to the similarity of their temporal pathways of change. Ten processes are uncovered ranging from various forms of suburbanization to revitalization and a suburban densification trend in the newest, rapidly growing cities in the study. Finally, we calculated a join-count statistic to quantify the observed spatial pattern of each of these neighbourhood types in each city.

The Dynamics of Urban Neighborhoods: A Survey of Approaches for Modeling Socio-Spatial Structure

“…One vibrant strand of work is found in analyzing the sequences using the optimal matching algorithm (Abbott, 1995;Gauthier et al, 2010). In these analyses, an initial geodemographic analysis is adopted to segment underlying urban space into specific neighborhood classes (Mikelbank, 2011;Delmelle and Thill, 2014;Delmelle, 2015Delmelle, , 2016Ling and Delmelle, 2016;Delmelle, 2017). Then, the historical experience of an urban area can be examined by summarizing the sequences of the identified neighborhood classifications it experiences.…”

Neighborhood Dynamics with Unharmonized Longitudinal Data

“…2008; Thomas et al. 2012), or, to a lesser extent, Self‐Organizing Maps (Arribas‐Bel and Schmidt 2013; Ling and Delmelle 2016; Delmelle 2017).…”

Extracting Interesting Regions and Trips from Taxi Trajectory Data

“…Kohonen's Self Organizing Map (SOM) (Kohonen, 1982) is a competitive Artificial Neural Network (ANN) that is classified as an unsupervised machine learning technique. Many trajectory data mining approaches such as (Ling and Delmelle, 2016;Shukla et al, 2012;Chen et al, 2008;Schreck et al, 2008) have used SOM for mining trajectory data. As far as this research is concerned, SOM was chosen as the primary clustering technique because it is an unsupervised learning technique which suits the nature of the trajectory data.…”