Summer school  teaching unit:
Spatial Data Analysis of Socio-Economic Units

Gilberto Camara– April 4, 2003.

Hours

Session 1: June 9: 90 min. + 90 min. (afternoon)

Content

This unit will present an introduction to spatial data analysis of areas, with the following topics:

  1. General topics: Spatial dependence and spatial autocorrelation.
  2. Analysis of Areas I: The modifiable area unit problem.
  3. Analysis of Areas II: Global and Local autocorrelation indexes.
  4. Analysis of Areas III: Spatial regression.
  5. Analysis of Areas IV: Analysis of small areas.

The classes will include a theoretical overview and a practical demonstration on the lab using a PC.

In this teaching unit, we will concentrate on areal units with counts and aggregated rates. In most cases, this means data associated to population surveys, like census and health statistics, and that are originally referred to individuals situated in specific points in space. For confidentiality reasons these data are aggregated in analysis units, usually delimited by closed polygons (census tracts, postal addressing zones, municipalities).

In this type of modeling we consider that the geographic space under study, region A, is a fixed set of spatial units. The most used model of distribution considers a stochastic process {Zi:i=1,2,…,n}, composed of a set of random variables. We seek to construct an approximation of the joint distribution of these variables Z={Z1,…,Zn}, where each random variable is associated to one of the areas and has a distribution to be estimated.

This tutorial will provide a primer on the main different types of spatial analysis of areas, considering aspects such as: (a) what happens when we change the scale of the spatial measurements? Does this imply an important difference in the results we obtain? (b) What are the limitations and possible errors when we use classical regression techniques for spatial data? What are the alternatives? (c) What happens when we calculate and compare rates and proportions on areal units of very different sizes and populations? What care must be taken when dealing with small areas?


Goals

The tutorial aims at providing the students with a general overview of the main areas of spatial data analysis. They should be able to understand the implications of using spatial data in the statistical inference process and to learn what information is possible to obtain from socio-economical units.

Students’ preparation in advance

Tutorials prepared for this teaching unit:

Spatial data analysis: a primer, by G. Câmara, A. Monteiro, S. Druck. M.Carvalho.

A tutorial on spatial data analysis of areas, by  G. Câmara, M.Carvalho.

References:

Bailey, T.; Gatrell, A. "Interactive Spatial Data Analysis". London, Longman Scientific and Technical, 1995.

Rees, Martin, Williamson, The Census Data System. Wiley, 2002.

P. Longley and M. Batty (eds) Spatial analysis: modelling in a GIS environment (Cambridge: Geoinformation International), 1997.

Fotheringham, A.S., Brunsdon, C., and Charlton, M.E., 2002, Geographically Weighted Regression: The Analysis of Spatially Varying Relationships, Chichester: Wiley.

Useful site:

AI-GEOSTATS: The Central Server for GIS & Spatial Statistics on the Internet. Hundreds of links on spatial analysis.

Teaching materials in Portuguese:

Gilberto Camara’s website.

Software downloads or URLs

GeoDa.   Software for Exploratory Spatial Data Analysis (Luc Anselin and co-workers - Free).

Space Stat. Software for Spatial Data Analysis (Luc Anselin).

ClusterSeer. Software para clustering of spatial data.

SPRING. Free GIS software by INPE, includes geostatistical tools and exploratory spatial data analysis.

TerraView. Open source software by INPE, includes tools for exploratory spatial data analysis.