Exploratory Spatial Data Analysis Using GeoDA : An ... · Prepared by Professor Ravi K. Sharma,...
Transcript of Exploratory Spatial Data Analysis Using GeoDA : An ... · Prepared by Professor Ravi K. Sharma,...
Exploratory Spatial Data Exploratory Spatial Data
Analysis Using Analysis Using GeoDAGeoDA: An : An
IntroductionIntroduction
Prepared by Professor Ravi K. Sharma, University of PittsburghModified for NBDPN 2007 Conference Presentation byProfessor Russell S. Kirby, University of Alabama at Birmingham
ObjectivesObjectives
Using MACDP data on chromosomal Using MACDP data on chromosomal
abnormalities measured across census abnormalities measured across census
tracts, we will demonstrate the use of tracts, we will demonstrate the use of
GeoDaGeoDa to:to:
�� Start a project, import data, use basic functionsStart a project, import data, use basic functions
�� Perform Exploratory Spatial Data Analysis (ESDA) Perform Exploratory Spatial Data Analysis (ESDA)
�� Calculate rates and weightsCalculate rates and weights
�� Create spatial weight matrixCreate spatial weight matrix
�� Perform spatial autocorrelationPerform spatial autocorrelation
The examples that follow are based on a dataset The examples that follow are based on a dataset
for countyfor county--level analysis of low birth weight for level analysis of low birth weight for
the state of Pennsylvania.the state of Pennsylvania.
GeoDaGeoDa
�� GeoDaGeoDa is a freely available software is a freely available software
program for exploratory spatial data program for exploratory spatial data
analysis (ESDA), developed by analysis (ESDA), developed by
Professor Luc Professor Luc AnselinAnselin of the of the
University of IllinoisUniversity of Illinois
�� It can be downloaded from the It can be downloaded from the
following URL:following URL:
•• https://https://www.geoda.uiuc.eduwww.geoda.uiuc.edu//
Beginning a ProjectBeginning a Project
Opening Opening GeoDaGeoDa
�� To begin click on the To begin click on the GeoDaGeoDa IconIcon
Starting a projectStarting a project
Open a map with shape fileOpen a map with shape file
The base mapThe base map
Editing featuresEditing features
Menu toolbar featuresMenu toolbar features
Icon toolbar featuresIcon toolbar features
Creating Maps and Creating Maps and
Selecting FeaturesSelecting Features
Create Create choroplethchoropleth mapsmaps
ChoroplethChoropleth map stepsmap steps
Creating Creating choroplethchoropleth ((quantilequantile) maps) maps
Creating Creating quantilequantile mapsmaps
Final Final choroplethchoropleth mapmap
Open a new copy of base mapOpen a new copy of base map
Create a new Create a new choroplethchoropleth mapmap
Dynamic map selection optionDynamic map selection option
Selecting map areasSelecting map areas
Table featuresTable features
Table sorting featuresTable sorting features
Specific table selectionSpecific table selection
Creating new variablesCreating new variables
Creating shape files from mapCreating shape files from map
Polygon to point shape filePolygon to point shape file
Create Create centroidscentroids for point filesfor point files
Exploratory Data Exploratory Data
Analysis (EDA)Analysis (EDA)
EDA: plotsEDA: plots
Variable selection for plotsVariable selection for plots
Plots: HistogramPlots: Histogram
Linkage: selecting featuresLinkage: selecting features
Linkage: selecting features (Linkage: selecting features (concon’’tt.).)
Generate and interpret box plotsGenerate and interpret box plots
Calculating RatesCalculating Rates
Create raw ratesCreate raw rates
Selecting variables for ratesSelecting variables for rates
Raw rates: by percentRaw rates: by percent
Saving ratesSaving rates
Identifying outliers from box plotsIdentifying outliers from box plots
Create excess risk ratesCreate excess risk rates
Excess risk mapExcess risk map
Creating Empirical Creating Empirical BayesBayes smoothingsmoothing
Map generated by EB smoothingMap generated by EB smoothing
Creating Weights: Creating Weights:
Examining spatial Examining spatial
relationshipsrelationships
Creating weightsCreating weights
Loading weight filesLoading weight files
Creating spatial ratesCreating spatial rates
Spatially smoothed mapSpatially smoothed map
Create weights: RookCreate weights: Rook
Text file of Rook weightsText file of Rook weights
Compare weights with map and tableCompare weights with map and table
View weight characteristicsView weight characteristics
Multiple views: weight histogram, Multiple views: weight histogram,
map and tablemap and table
Create weights: QueenCreate weights: Queen
Compare multiple featuresCompare multiple features
Creating weights: neighborsCreating weights: neighbors
Reviewing weightsReviewing weights
Creating weights: nearest neighborsCreating weights: nearest neighbors
Histogram of neighbor weightsHistogram of neighbor weights
Autocorrelation: Autocorrelation:
Identifying clustersIdentifying clusters
Global Moran Global Moran
�� It is a measure of spatial autocorrelation It is a measure of spatial autocorrelation (feature similarity) based not only on (feature similarity) based not only on feature locations or attribute values alone feature locations or attribute values alone but also on both feature locations and but also on both feature locations and feature values simultaneously. feature values simultaneously.
�� Given a set of features and an associated Given a set of features and an associated attribute, it evaluates whether the pattern attribute, it evaluates whether the pattern expressed is clustered, dispersed, or expressed is clustered, dispersed, or random. random.
�� A Moran's Index value near +1.0 indicates A Moran's Index value near +1.0 indicates clustering; an index value near clustering; an index value near --1.0 1.0 indicates dispersionindicates dispersion
Global MoranGlobal Moran
Autocorrelation: weight file requiredAutocorrelation: weight file required
Global Moran resultGlobal Moran result
Randomization featureRandomization feature
Randomization: graph resultRandomization: graph result
Randomization: Envelope SlopesRandomization: Envelope Slopes
Local Moran (LISA)Local Moran (LISA)
�� The local Moran test (The local Moran test (AnselinAnselin 19951995), detects local ), detects local spatial autocorrelation. It can be used to identify spatial autocorrelation. It can be used to identify local clusters (regions where adjacent areas have local clusters (regions where adjacent areas have similar values) or spatial outliers (areas distinct similar values) or spatial outliers (areas distinct from their neighbors).from their neighbors).
�� The Local Moran statistic decomposes Moran's I The Local Moran statistic decomposes Moran's I ((Moran 1950Moran 1950) into contributions for each location, ) into contributions for each location, IiIi. The sum of . The sum of Ii Ii for all observations is for all observations is proportional to Moran's I, an indicator of global proportional to Moran's I, an indicator of global pattern. pattern.
�� Thus, there can be two interpretations of Local Thus, there can be two interpretations of Local Moran statistics, as indicators of local spatial Moran statistics, as indicators of local spatial clusters and as a diagnostic for outliers in global clusters and as a diagnostic for outliers in global spatial patterns.spatial patterns.
Local MoranLocal Moran
LISA: Significance mapLISA: Significance map
LISA: Cluster mapLISA: Cluster map