Spatial Clustering Yogi Vidyattama
Main methodology Local Moran Index of Spatial Autocorrelation where: Zi = deviation of point i to the mean Wij = the spatial weights matrix m2 = total variance ( ) Weight matrix The nature of spatial relationship: contiguity
Methodology Weight matrices Describes the nature of the spatial relationships Rook-Contiguity based As opposed to Queen-Contiguity based Distance based
What it has been used for Concentration area of disadvantage Joblessness Overcrowded housing Less developed region Ethnic group: Ancestry, Language Why? Spill over effect Intertemporal effect Public service/ infrastructure distribution Further analysis necessary Changes over time Different behaviour/attitude Impact on regression
Children in Jobless household
Children in Jobless household
Children in Jobless household
Children in Jobless household
SYDNEY Ethnicity
MELBOURNE Ethnicity
Cluster of HDI 1999
Cluster of HDI 2008
Moran’s “Arrow” plot of HDI 1999-2008
Other used Disease Crime Natural disaster and its impact Contagion, connected topography Crime Vulnerable target, location of criminal Natural disaster and its impact Flood, fire, earthquake
Software OpenGeoDa 09.09.12 AURIN PORTAL
Step in AURIN portal Choose or import your dataset Spatialise your dataset Create your weight matrix
Step in AURIN portal (2) Calculate the Local Moran’s I statistics Visualise the result
Analyse in AURIN portal Percentage is mainly used but comparing the result to the result in number is often important