Conceptualization of Spatial

Relationships
Spatial Data Analysis
 Spatial analysis
Spatial analysis is a collection of methods,
statistics and techniques, which integrates
concepts such as location, area, distance,
interaction to analyze, investigate and explain in
a geographic context, patterns, actions or
behaviors among spatially referenced
observations that arise as a result of a process
operating in space.
 Geospatial analysis
• Geospatial analysis is the collection of spatial
analysis methods, techniques and models that
are integrated into geographic information
systems (GIS).
Conceptualization of spatial relationships

Conceptualization of spatial relationships is the

modeling of the relationships and interactions
between features across space..
 Purpose
• Conceptualizing of spatial relationships is used
to define what is to be regarded as close, far,
adjacent or neighboring and is essential prior
to any geographical analysis and spatial
statistical tool implementation.
 spatial statistics tools
• Many spatial statistics tools require that a conceptualization
method of spatial relationships be set prior to any analysis.
Some of these statistics are as follows:
• Global spatial autocorrelation (Global Moran’s I, General G-
Statistic)
• Cluster and outlier analysis (Anselin Local Moran’s I)
• Hot Spot Analysis (Getis-Ord Gi)
• Spatiotemporal Autocorrelations (Bivariate Moran’s I,
Differential Moran’s I)
• Generate spatial weights matrix
• Geographically weighted regression
 Distance
• Define distance threshold or distance
function:
• Distance type (e.g., Euclidean, Manhattan,
Minkowski, Network)
• Fixed distance band (distance threshold)
• Distance decay (distance function)
 Distance Measurement
Among the most common distance measures
used in geographical analysis are the Euclidean
distance, the Manhattan distance, the
Minkowski distance, the Pearson’s correlation
distance, the Spearman correlation distance, the
network distance, and the geodetic distance. In
spatial statistics, Euclidean and Manhattan
distance are those most widely used.
 Definitions and Formulas
• For two points A, B, where X1, Y1 are the
coordinates of point A and X2, Y2 are the
coordinates of point B, measured on a
projected coordinate system (plane surface
using Cartesian coordinates), distance can be
defined as follows:
• Euclidean distance is the distance between
two points A and B connected by a straight
line calculated as
 Definitions and Formulas
• Manhattan distance is the vertical plus
horizontal difference (measured along the
axes) between points A and B and is calculated
as

S = |X2 - X1| + |Y2 - Y1|

 Minkowski distance
• Minkowski distance is a generalization of
Euclidean and Manhattan distance:
S = ((X2 X1) p + (Y2 Y1) p ) 1/p

For p = 1, we obtain the Manhattan distance and for p = 2, the

Euclidean distance. Minkowski distance is used among others in
Principal Component Analysis and Clustering Analysis
 Pearson’s correlation distance
• Pearson’s correlation distance considers two
vectors to be close (similar) if they are highly
correlated. In a multivariate dataset and for
two variables with correlation r, Pearson’s
correlation distance is for correlation
coefficient r definition and equation):
• S=1-r
 distance decay
• The range of distance is closely related to the
scale of analysis. Two widely used methods to
define the range of distance are the fixed
distance band and the distance decay
 Distance Decay
• Distance Decay Definition Distance decay is
any function that implies a continuous,
smooth and attenuating effect of distance on
the attribute values of neighboring spatial
entitie
 Fixed Distance Band
• Fixed Distance Band (Sphere of Influence)
Definition Fixed distance band is a distance
value expressing the size of a sphere of
influence around a spatial object. All spatial
entities inside the sphere of influence are
weighted equally, while spatial entities outside
the zone are assigned zero weight
 Adjacency.
• Define which objects are regarded as adjacent
• Contiguity edges only (Rook’s Case)
• Contiguity edges corners (Queen’s Case)
• Higher-order contiguity
• Interaction (this includes distance and
adjacency
 Contiguity
Contiguity is a spatial property that describes whether a target object and
one or more other objects are in close proximity. In practice, contiguity
refers to which polygons are assigned as neighbors for a single target
object.
The most common contiguity conceptualization methods for polygon
features
are:
Contiguity edges only (Rook’s Case). Only those polygons that share a
common border (edge) are regarded as neighboring and are included in
calculations for the target polygon
Contiguity edges corners (Queen’s Case). In this case, polygons that
share borders and also have common corners (nodes) are considered
neighbors and are included in calculations for the target polygon
 Use of Contiguity
• Contiguity is used to define neighborhoods
used in the calculation of the spatial weights
matrix and various spatial statistics
 Adjacency Matrix
• In the context of spatial analysis and for
polygon representation, the adjacency matrix
is a square matrix, the elements of which
indicate whether pairs of polygons are
adjacent or not.
• An adjacency matrix is used to represent the
various forms of contiguity and to define
neighborhoods for further spatial analysis
 Adjacency Matrix
• An adjacency matrix is a symmetrical matrix the off-
diagonal elements of which take values of either 0 or 1
and the diagonal elements of which have no values.
• If two spatial entities are adjacent (either first-order or
higher-order), then the corresponding matrix element is
set to 1 and 0 otherwise.
• For a given target polygon in the matrix, the polygons with
an adjacency value of 1 (in the same row or column)
define its neighborhood (the set of features to be taken
into account for calculating spatial statistic for the target
feature).
 Interaction
• Interaction is the degree of linkage between
two locations, the origin and the destination.
It is calculated as a combination of distance
and adjacency.
• In spatial statistical analysis, calculating
interaction can be used as a way to calculate
weights in a spatial weights matrix
• In its simplest form, its formula is:
 Neighborhood.
• Define what makes a neighborhood
• k-nearest neighbors
• Proximity polygons
• Delaunay triangulation
 Neighborhood and Neighbors
Neighborhood in the spatial analysis context is a
geographically localized area to which local spatial
analysis and statistics are applied based on the
hypothesis that objects within the neighborhood are
likely to interact more than those outside it.
Most of these statistics require a neighborhood
definition and the construction of a spatial weights
matrix that reflects the intensity of the relationships
among the spatial entities in this neighborhood
 k-Nearest Neighbors (k-NN
In the context of spatial analysis, k-NN is a method used
to define a neighborhood based on the k-nearest
neighbors to the target object.
This method ensures that each object will have at least a
certain number of neighbors. More broadly, it is used to
Define the neighborhood (region) in which spatial
objects will be accounted in spatial statistics calculations
Define the neighborhood in which spatial objects are
likely to be more similar than objects at further distances
Model spatial relationships
 Space–Time.
• Define distance and time windows for
spatiotemporal analysis
 Space–Time.
• Space–time window is a conceptualization
method for defining a neighborhood based on
both distance and time. An object is in the
same neighborhood with the target object if it
falls within the specified distance and also falls
within the specified time interval.
• This method can be used to identify
spatiotemporal hot spots or clusters created
based on spatial and temporal proximity.
 Proximity Polygons
• Proximity polygons are polygons that divide
space into regions so that the nearest centroid
of each point in a region is the one in the
polygon it lies in.
• Proximity polygons are also called Thiessen
polygons or Voronoi polygons
 Delaunay Triangulation and Triangular
Irregular Networks (TIN
• Delaunay triangulation partitions space by creating
triangles from point features or polygon centroids
whose proximity polygons share an edge
• This method is suitable in cases where isolated
polygons (e.g., islands) exist in the dataset or
where the spatial distribution of objects is
abnormal
• Points or centroids connected by triangle edges
are regarded as neighbors. This ensures that each
object will have at least one neighbor
 Spatial Weights and Row Standardization
• Spatial weights are numbers that reflect some sort of
distance, time or cost between a target spatial object and
every other object in the dataset or specified neighborhood.
Spatial weights quantify the spatial or spatiotemporal
relationships among the spatial features of a neighborhood.
• Spatial weights matrix is the matrix that stores the spatial
weights.
• Row standardization is the process of scaling the spatial
weights to a range between 0 and 1. It is used to avoid
biased data sampling or when data are aggregated from
larger datasets