Exploratory Spatial Data Analysis Using GeoDA: : An Introduction Prepared by Professor Ravi K. Sharma, University of Pittsburgh Modified for NBDPN 2007 Conference Presentation by Professor Russell S. Kirby, University of Alabama at Birmingham
Objectives Using MACDP data on chromosomal abnormalities measured across census tracts, we will demonstrate the use of GeoDa to: Start a project, import data, use basic functions Perform Exploratory Spatial Data Analysis (ESDA) Calculate rates and weights Create spatial weight matrix Perform spatial autocorrelation The examples that follow are based on a dataset for county-level analysis of low birth weight for the state of Pennsylvania.
GeoDa GeoDa is a freely available software program for exploratory spatial data analysis (ESDA), developed by Professor Luc Anselin of the University of Illinois It can be downloaded from the following URL: https://www.geoda.uiuc.edu www.geoda.uiuc.edu/
Beginning a Project
Opening GeoDa To begin click on the GeoDa Icon
Starting a project
Open a map with shape file
The base map
Editing features
Menu toolbar features
Icon toolbar features
Creating Maps and Selecting Features
Create choropleth maps
Choropleth map steps
Creating choropleth (quantile)) maps
Creating quantile maps
Final choropleth map
Open a new copy of base map
Create a new choropleth map
Dynamic map selection option
Selecting map areas
Table features
Table sorting features
Specific table selection
Creating new variables
Creating shape files from map
Polygon to point shape file
Create centroids for point files
Exploratory Data Analysis (EDA)
EDA: plots
Variable selection for plots
Plots: Histogram
Linkage: selecting features
Linkage: selecting features (con( con t.)
Generate and interpret box plots
Calculating Rates
Create raw rates
Selecting variables for rates
Raw rates: by percent
Saving rates
Identifying outliers from box plots
Create excess risk rates
Excess risk map
Creating Empirical Bayes smoothing
Map generated by EB smoothing
Creating Weights: Examining spatial relationships
Creating weights
Loading weight files
Creating spatial rates
Spatially smoothed map
Create weights: Rook
Text file of Rook weights
Compare weights with map and table
View weight characteristics
Multiple views: weight histogram, map and table
Create weights: Queen
Compare multiple features
Creating weights: neighbors
Reviewing weights
Creating weights: nearest neighbors
Histogram of neighbor weights
Autocorrelation: Identifying clusters
Global Moran It is a measure of spatial autocorrelation (feature similarity) based not only on feature locations or attribute values alone but also on both feature locations and feature values simultaneously. Given a set of features and an associated attribute, it evaluates whether the pattern expressed is clustered, dispersed, or random. A Moran's Index value near +1.0 indicates clustering; an index value near -1.0 indicates dispersion
Global Moran
Autocorrelation: weight file required
Global Moran result
Randomization feature
Randomization: graph result
Randomization: Envelope Slopes
Local Moran (LISA) The local Moran test (Anselin( 1995), detects local spatial autocorrelation. It can be used to identify local clusters (regions where adjacent areas have similar values) or spatial outliers (areas distinct from their neighbors). The Local Moran statistic decomposes Moran's I (Moran 1950) ) into contributions for each location, Ii.. The sum of Ii for all observations is proportional to Moran's I, an indicator of global pattern. Thus, there can be two interpretations of Local Moran statistics, as indicators of local spatial clusters and as a diagnostic for outliers in global spatial patterns.
Local Moran
LISA: Significance map
LISA: Cluster map