Mapping the `DNA`

MAPPING THE ‘DNA’ OF URBAN
NEIGHBORHOODS: CLUSTERING
LONGITUDINAL SEQUENCES OF
NEIGHBORHOOD SOCIOECONOMIC
CHANGE
Dr. Elizabeth C. Delmelle
University of North Carolina at Charlotte
[email protected]
A LITTLE HISTORY: URBAN GEOG 101
Burgess (1925)
Hoyt (1939)
http://www.martinsaphug.com/learn/units/urbanization/urban-structure-models/
Harris & Ullman (1945)
HISTORY LESSON CONT.
Spatial Dynamics
Downgrading
• ‘Invasion or succession’
• Housing Age
• Commuting cost vs. Lot
size.
Decline
Revitalization
Wealthy
‘Inner-Ring’
Decline
EMPIRICAL ANALYSES: THE STATE OF
THE ART
1. What are the most common pathways of neighborhood change
through time?
Longitudinal Sequences
Van Criekingen and Decroly (2003)
Upgrading Processes: Gentrification,
Marginal Gentrification, Upgrading,
Incumbent upgrading
Wei and Knox (2014)
EMPIRICAL ANALYSES: THE STATE OF
THE ART
1. What
are the spatial patterns of neighborhood dynamics?
Change Maps (Difference between two time periods)
Tierney & Petty (2015)
EMPIRICAL ANALYSES: THE STATE OF
THE ART
1. What
are the spatial patterns of neighborhood dynamics?
Cross-Sectional Maps
OTHER TECHNIQUES
Latent Growth Curve Models
Delmelle et al. (2013)
Seguin et al., 2012
Self-Organizing Map Trajectories
RESEARCH QUESTIONS
1. What are the most common pathways of neighborhood
change through time?
2. Which neighborhoods have followed similar trajectories and
where are they located?
OR: What are the spatial patterns of neighborhood dynamics?
PROPOSED METHODOLOGY
Cluster Analysis
(k-means)
Normalize
Variables
Sequential
Pattern Mining
Socioeconomic
DATA
Longitudinal Tract Database
Brown University
Aligned 1970-2010 Census Tract
Data to 2010 Census Tract
Boundaries
Case Study Area: Census
Tracts in Cook County, IL
(Chicago) n=1318
% persons with at least a 4-year degree
% Unemployed
% Manufacturing Employees
% Below Poverty Level
Housing
% Owner Occupied
% Multiunit Structures
Median Home Value
% Structures built more than 30 years ago
% Household heads moved into unit < 10 years
ago
% Vacant Housing
Demographic
% Persons age 60+
% Persons < 18
K-MEANS CLASSES
Suburban
•
•
•
•
•
High homeownership
Above average college education
Lots of children
New housing Stock
Large share of “in-movers”
Stability
•
•
•
•
•
Older Population composition
Lived in homes for a long time
Low Poverty
Low Unemployment
High Homeownership
Blue Collar
• Largest share of blue collar workers
• Oldest Housing Stock
• Home values below city mean
CLASSES
Struggling
• Highest poverty rates
• High unemployment
• College education levels far
below city mean
• Vacant Housing
New Starts
•
•
•
•
•
Highest educated residents
Highest home values
Multi-unit structures
Few children
Lots of recent in-movers
SEQUENTIAL PATTERN MINING
Measure
Sequence
Similarity
Cluster
Sequences
Optimal Matching Algorithm
o Cost of Inserting, Deleting, or Substituting elements in
one sequence in order to completely transform it into
another sequence.
A: {Suburban, Suburban, Suburban, Stability, Blue Collar}
B: {Suburban, Suburban, Suburban, Blue Collar, Blue Collar}
Map Cluster
Membership
TraMineR Package
SEQUENTIAL PATTERN MINING CONT.
Transition Substitution Costs
P(i|j)
Suburban Stability
Blue
Struggling New
Collar
Starts
Suburban 0
1.68
1.92
1.99
1.94
Stability
1.68
0
1.83
1.99
1.93
Blue
1.90
1.83
0
1.78
1.87
Struggling 1.99
1.99
1.78
0
1.90
New
1.93
1.87
1.90
0
Collar
Starts
1.94
LONGITUDINAL
CLUSTERS 1-4
(1) Upgrading from Struggling
• Struggling  New Starts (Gentrification)
• Struggling  Blue Collar (Marginal
Gentrification)
(2) Blue Collar to New Starts
• 2nd Upgrading process
(3) Stable New Starts
• Highest Real Estate values, consistent flow of
recent, highly educated residents.
(4) Stability
• Largest Cluster
• Stable, middle-class suburbs
LONGITUDINAL
CLUSTERS 5-8
(5) Stability to Blue Collar
• Downgrading
• Spatial frontier of suburban decline
(6) Persistently Struggling
(7) Stable Blue Collar
(8) Blue Collar to Struggling
• Smallest cluster
• Less spatially compact
LONGITUDINAL
CLUSTERS 9 & 10
(9) Suburban to Stability
• Aging of Suburbs
(10) Suburban
CHICAGO SCHOOL PATTERNS
REVISITED
o Wealthy urban along banks
of lake (Hoyt); Suburban
wealthy outer rings of county
(Burgess).
oConcentric Rings
oStable Blue Collar &
Struggling in ‘Multiple Nuclei’
(Harris & Ullman)
Spatial Contours of
Urban Dynamics
ANOTHER PERSPECTIVE
DELVING INTO SOME EXPLANATIONS
o Clusters 7 & 8:
o Both Situated in transformative middle ring
o Both Started off as Blue Collar.
o Cluster 7 remained BC, Cluster 8 declined to Struggling
o Clusters 1 & 6:
o Upgraded from Struggling vs. Persistently Struggling
Cluster 7: Doubled
Share of Hispanic
Residents (30 to 60%)
Cluster 8: Doubled
Share of black Residents
(27 to 66%)
Cluster 1: Racial
diversity in 1970 (50%
black, 47% white)
Cluster 6: From 78 to
89% black, 19702010.
CONCLUSIONS & THE FUTURE
The Promising
Method produced intuitive results. 10 clusters.
 5 Stability
 3 Downgrading
 2 Upgrading
Illustrated durability of Chicago-School Spatial Patterns (from a dynamic
perspective).
Discerns transition zones. Could be helpful for policy makers\the public..
CONCLUSIONS & THE FUTURE
The But…
o I picked an easy city…
Charlotte’s DNA
THE FUTURE
o Expand to other cities.
o A real explanatory model.
o Alternative neighborhood
classifications – geodemographic
segmentation data?
THANK YOU.
Questions?
REFERENCES
Delmelle, E.C., Thill, J-C., Furuseth, O., Ludden, T. (2013) Trajectories of multidimensional neighborhood
quality of life change. Urban Studies 50, 923-941.
Seguin, A-M., Apparicio, P., Riva, M. (2012) Identifying, mapping, and modelling trajectories of poverty
at the neighborhood level: The case of Montreal, 1986-2006. Applied Geography 35, 265-274.
Teernstra, A.B., & Van Gent, W.P.C. (2012) Puzzling patterns in neighborhood change: Upgrading and
downgrading in highly regulated urban housing markets. Urban Geography 33, 91-119.
Tierney, S. & Petty, C. (2014) Gentrification in the American heartland? Evidence from Oklahoma City.
Urban Geography
Van Criekingen, M. & J-M Decroly (2003) Revisiting the diversity of gentrification: neighborhood
renewal processes in Brussels and Montreal. Urban Studies 40, 2451-2468.
Wei, F., & Knox, P. (2014) Neighborhood change in Metropolitan America, 1990-2010. Urban Affairs
Review 50, 459-489.