Select the "Bar Charts" option from the dropdown menu, then click on one neighborhood to view its standardized (z-score) values for 18 variables from 1980-2010.
A total of 18 bar charts will be displayed on the right-hand pane.
A red bar indicates a positive value (greater than the metropolitan average of that year), while the blue bar represents a negative value (below the metropolitan average for that year).
Hover over each bar to see an interpretation of the value.
Note: each variable is normalized relative to its own metropolitan area using z-score. For example, home values in the Charlotte MSA for 2000 are relative to the mean of that MSA for that year and not compared to the absolute value of home values in New York City.
Here the z-value, also referred to as a standardardized score, is a dimensionless quantity obtained by subtracting the metropolitan area mean from an individual neighborhood value and then dividing the difference by the metropolitan-wide standard deviation.
It is used to put all values on the same numerical scale where the average value is equal to 0, values above the mean are positive, and values below the mean are negative.
SOM is a data visualization method that projects multidimensional data onto a 2-dimensional output space.
It is an artificial neural network-based technique that arranges the data in such a way that observations (in this case neighborhoods) that are most similar to each other across the set of input variables are located in close proximity to one another on the output space.
In this case, all data from all four decades (1080, 1990, 2000, 2010) are entered into the SOM procedure at once so that a neighborhood¡¯s position on the output space can be traced, following its temporal trajectory.
The SOM output space is segmented into 9 groups using a K-means clustering approach to delimit homogeneous groups of neighborhoods.
Select the "SOM" from the dropdown menu, then click on one neighborhood. An animated line follows that neighborhood¡¯s movement on the output space.
A white line denotes change from 1980 to 1990; a grey line from 1990 to 2000 and a dark line from 2000 to 2010. A larger trajectory generally depicts a greater amount of change.
A national typology of neighborhood sequences of change was constructed by classifying the sequences followed by each neighborhood between the 9 clusters.
That typology features 34 dominant sequences of change. A neighborhood¡¯s membership within one of those sequence groups is also indicated.
Thank the support from the Department of Geography & Earth Science and College of Liberal Arts & Sciences in University of North Carolina at Charlotte.
We are grateful to Alexander Marciniak and Elie Saliba from the University of North Carolina at Charlotte for providing web server support.