spatial regression · • spatial weights • exclude some interactions • constrain the number of...
TRANSCRIPT
![Page 1: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/1.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Luc Anselin
Spatial Regression1. Introduction and Review
http://spatial.uchicago.edu
![Page 2: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/2.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• matrix algebra basics
• spatial econometrics - definitions
• pitfalls of spatial analysis
• spatial autocorrelation
• spatial weights
![Page 3: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/3.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Matrix Algebra Basics
![Page 4: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/4.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Review (on your own)
• matrix multiplication, quadratic forms
• matrix inverse
• determinant
• Cholesky decomposition
• eigenvalues, eigenvalue decomposition
• optional: matrix partial derivates
![Page 5: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/5.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Spatial EconometricsDefinitions
![Page 6: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/6.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• When is Analysis Spatial?
• geo-spatial data: location + value (attribute)
• “non-spatial” analysis:location does NOT matter = locational invariance
• spatial analysis:when the location changes, the information content of the data changes
![Page 7: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/7.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Spatial Distribution
![Page 8: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/8.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
A-Spatial DistributionHistogram
![Page 9: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/9.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Spatial AnalysisGlobal Spatial Autocorrelation
Moran Scatter Plot
![Page 10: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/10.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Definition of Spatial Econometrics (e.g. , Anselin 1988, 2006)
• a subset of econometric methods concerned with spatial aspects present in cross-sectional and space-time observations
• explicit treatment of location, distance and arrangement in model specification, estimation, diagnostics and prediction
• spatial effects: dependence, heterogeneity
![Page 11: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/11.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
spatial econometrics
![Page 12: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/12.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Spatial Effects
• spatial heterogeneity
• structural change, varying coefficients
• standard methods apply
• spatial dependence
• two-dimensional and multidirectional cross-sectional dependence
• not a direct extension of time series methods
![Page 13: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/13.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Four dimensions of spatial econometrics
• specifying the structure of spatial dependence/heterogeneity
• testing for the presence of spatial effects
• estimating models with spatial effects
• spatial prediction
![Page 14: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/14.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Pitfalls of Spatial Analysis
![Page 15: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/15.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Ecological Fallacy
• individual behavior cannot be explained at the aggregate level
• issue of interpretation
• e.g., county homicide rates do not explain individual criminal behavior
•model aggregate dependent variables with aggregate explanatory variables
• alternative: multilevel modeling
![Page 16: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/16.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Modifiable Areal Unit Problem (MAUP)
• what is the proper spatial scale of analysis?
• a million spatial autocorrelation coefficients (Openshaw)
• spatial heterogeneity - different processes at different locations/scales
• both size and spatial arrangement of spatial units matter
![Page 17: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/17.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Source: Scholl and Brenner (2012)
![Page 18: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/18.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Change of Support Problem (COSP)
• variables measured at different spatial scales
• nested, hierarchical structures
• non-nested, overlapping
• solutions
• aggregate up to a common scale
• interpolate/impute - Bayesian approach
![Page 19: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/19.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Spatial Autocorrelation
![Page 20: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/20.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• The Null Hypothesis
• spatial randomness is absence of any pattern
• spatial randomness is not very interesting
• if rejected, then there is evidence of spatial structure
![Page 21: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/21.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Tobler’s First Law of Geography
• everything depends on everything else, but closer things more so
• structures spatial dependence
• importance of distance decay
![Page 22: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/22.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Positive Spatial Autocorrelation
• impression of clustering
• clumps of like values
• like values can be either high (hot spots) or low (cold spots)
• difficult to rely on human perception
![Page 23: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/23.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
< positive s.a.
random >
![Page 24: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/24.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Negative Spatial Autocorrelation
• checkerboard pattern
• hard to distinguish from spatial randomness
![Page 25: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/25.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
< negative s.a.
random >
![Page 26: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/26.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Spatial Autocorrelation Statistic
• captures both attribute similarity and locational similarity
• how to construct an index from the data that captures both attribute similarity and locational similarity (i.e., neighbors are alike)
![Page 27: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/27.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Attribute Similarity
• summary of the similarity (or dissimilarity) of observations for a variable at different locations
• variable y
• locations i, j
• how to construct f(yi, yj)
• cross product, yi.yj
• squared difference, (yi -yj)2
• absolute difference, |yi - yj|
![Page 28: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/28.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Locational Similarity
• formalizing the notion of neighbors = spatial weights (wij)
• when are two spatial units i and j a priori likely to interact
• not necessarily a geographical notion, can be based on social network concepts or general distance concepts (distance in multivariate space)
![Page 29: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/29.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• General Spatial Autocorrelation Statistic
• general form
• sum over all observations of an attribute similarity measure with the neighbors
• f(xixj) is attribute similarity between i and j for x
• wij is a spatial weight between i and j
• statistic = Σij f(xixj).wij
![Page 30: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/30.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Moran’s I
• the most commonly used of many spatial autocorrelation statistics
• I = [ Σi Σj wij zi.zj/S0 ]/[Σi zi2 / N]
• with zi = yi - mx : deviations from mean
• cross product statistic (zi.zj) similar to a correlation coefficient
• value depends on weights (wij)
![Page 31: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/31.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Spatial Weights
![Page 32: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/32.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Basic Concepts
![Page 33: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/33.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Why Spatial Weights
• formal expression of locational similarity
• spatial autocorrelation is about interaction
• n x (n - 1)/2 pairwise interactions but only n observations in a cross-section
• insufficient information to extract pattern of interaction from cross-section
• example: North Carolina has 100 counties5,000 pairwise interactions, 100 observations
![Page 34: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/34.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Solution
• impose structure
• limit the number of parameters to be estimated
• incidental parameter problem = number of parameters grows with sample size
• for spatial interaction, number of parameters grows with n2
![Page 35: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/35.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Spatial Weights
• exclude some interactions
• constrain the number of neighbors, e.g., only those locations that share a border
• single parameter = spatial autocorrelation coefficient
• strength of interaction = combined effect of coefficient and weights
• small coefficient with large weights
• large coefficient with small weights
![Page 36: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/36.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
six polygons - neighbors share common border
![Page 37: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/37.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
neighbor structure as a graph
nodelink
![Page 38: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/38.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Spatial Weights Matrix Definition
• N by N positive matrix W with elements wij
• wij non-zero for neighbors
• wij = 0, i and j are not neighbors
• wii = 0, no self-similarity
![Page 39: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/39.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Geography-Based Spatial Weights
![Page 40: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/40.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Binary Contiguity Weights
• contiguity = common border
• i and j share a border, then wij = 1
• i and j are not neighbors, then wij = 0
• weights are 0 or 1, hence binary
![Page 41: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/41.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
binary contiguity weights matrix for six-region example
![Page 42: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/42.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Copyright © 2011 by Luc Anselin, All Rights Reserved
1 2 3
4 5 6
7 8 9
Regular Grid Contiguity
contiguity on a regular grid - different definitions
![Page 43: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/43.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Copyright © 2011 by Luc Anselin, All Rights Reserved
1 2 3
4 5 6
7 8 9
Rook Contiguity
Neighbors for 5: 2, 4, 6, 8Common Border
rook contiguity - edges only2, 4, 6, 8 are neighbors of 5
![Page 44: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/44.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Copyright © 2011 by Luc Anselin, All Rights Reserved
1 2 3
4 5 6
7 8 9
Rook Contiguity
Neighbors for 5: 2, 4, 6, 8Common Border
bishop contiguity - corners only1, 3, 7, 9 are neighbors of 5
Copyright © 2011 by Luc Anselin, All Rights Reserved
1 2 3
4 5 6
7 8 9
Bishop Contiguity
Neighbors for 5: 1, 3, 7, 9Common Vertex
![Page 45: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/45.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Copyright © 2011 by Luc Anselin, All Rights Reserved
1 2 3
4 5 6
7 8 9
Rook Contiguity
Neighbors for 5: 2, 4, 6, 8Common Border
queen contiguity - edges and corners5 has eight neighbors
Copyright © 2011 by Luc Anselin, All Rights Reserved
1 2 3
4 5 6
7 8 9
Queen Contiguity
8 Neighbors for 5Both Border and Vertex
![Page 46: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/46.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Solano county, CA contiguity
rook
queen
![Page 47: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/47.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Distance-Based Weights
• distance between points
• distance between polygon centroids or central points
• in general, can be any function of distance that implies distance decay, e.g., inverse distance
• in practice, mostly based on a notion of contiguity defined by distance
![Page 48: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/48.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Distance-Band Weights
• wij nonzero for dij < dless than a critical distance d
• potential problem: isolates = no neighbors
• make sure critical distance is max-min, i.e., the largest of the nearest neighbor distance for each observation
![Page 49: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/49.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• k-Nearest Neighbor Weights
• k nearest observations, irrespective of distance
• fixes isolates problem for distance bands
• same number of neighbors for all observations
• in practice, potential problem with ties
• needs tie-breaking rule (random, include all)
![Page 50: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/50.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Spatial Weights Transformations
![Page 51: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/51.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Row-Standardized Weights
• rescale weights such that Σj wij = 1
• wij* = wij / Σj wij
• constrains parameter space
• makes analyses comparable
• spatial lag = average of the neighbors
![Page 52: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/52.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
row-standardized weights matrix
![Page 53: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/53.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Higher Order Weights
![Page 54: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/54.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
second order contiguity: neighbor of neighbor
![Page 55: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/55.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
redundancy in higher order contiguitypaths of lenght 2 between 1 and other cells
![Page 56: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/56.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Higher Order Weights
• recursive definition
• k-th order neighbor is first order neighbor of (k-1)th order neighbor
• avoid duplication, only unique neighbors of a given order (not both first and second order)
• pure contiguity or cumulative contiguity, i.e., lower order neighbors included in weights
![Page 57: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/57.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Solano county, CA, second order contiguity
exclusive of first order inclusive of first order
![Page 58: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/58.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
Properties of Weights
![Page 59: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/59.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Connectivity Histogram
• histogram of number of neighbors
• neighbor cardinality
• diagnostic for “isolates” or neighborless units
• assess characteristics of the distribution
![Page 60: Spatial Regression · • Spatial Weights • exclude some interactions • constrain the number of neighbors, e.g., only those locations that share a border • single parameter](https://reader034.vdocuments.us/reader034/viewer/2022042316/5f04accc7e708231d40f2567/html5/thumbnails/60.jpg)
Copyright © 2017 by Luc Anselin, All Rights Reserved
• Things to Watch for
• isolates
• need to be removed for proper spatial analysis
• do not need to be removed for standard analysis
• very large number of neighbors
• bimodal distribution