Chapter 2: The Pitfalls and Potential of Spatial Data

1
Chapter 2 The Pitfalls and Potential of Spatial
Data
(OSullivan and Unwin, 2003)
2
Bad News First

• Spatial Autocorrelation
• Data from locations near one another in space are
  more likely to be similar than data from
  locations remote from one another (positive,
  negative, zero)
• Modifiable Areal Unit Problem (MAUP)
• Ecological Fallacy
• Scale
• how an object type is represented is scale
  dependent
• Nonuniformity
• space is not uniform
• spatial distribution of people is bound to have
  clusters and gaps (doesnt conform to regular
  statistical methods)

3
Modifiable Areal Unit Problem

• This is a problem of data aggregation and the
  impact that the level of aggregation has on the
  patterns and relationships observed
• Usually, regression relationships are
  strengthened by aggregation

4
Ecological Fallacy

• High Correlation between crime and low-income
  areas

• Commit EF if we conclude that low-income people
  commit crimes
• Perhaps, low-income households dont have good
  alarm systems Or drug-addicts (regardless of
  income) live in these areas and commit crimes

5
Now for the Good News

• Spatial Analysis focuses on the distribution of
  data in space spatial relationships are
  described using the following
• Distance (dist. between spatial entities)
• Adjacency (spatial relationship between entities
  adjacent or not adjacent)
• Interaction (I 1/D) (the nearer things are,
  the more related they are)
• Neighborhood- region of space within a certain
  distance of a particular entity

6
Proximity Polygons
The proximity polygon of an entity is that region
of space that is closer to that entity than it is
to any other.
7
Variogram Cloud
Helps explore the spatial dependence between
attributes of objects plot the differences in
attribute values for pairs of entities against
the difference in their location (dist. between
entities). A variogram cloud is a tool used to
study spatial dependencies between attributes of
objects.
8
Questions

• What are the 5 major problems in the analysis of
  geographic information???
• How are proximity polygons constructed for point
  objects??