The application of flexible scan statistics in detecting poverty hotspots in Java Island

Abstract

Poverty hotspot detection is one of the study to know the set of area which has a high number of poverty. Previous study using Circular Spatial Scan Statistics found a problem which is low power to reject null hypothesis since it use circle window as the basis of scan. Flexible scan statistics provide better aproach by using a set of connected regions as window scan. Java Islands is the highest in number of people, with almost 60% of Indonesia people living in Java. By Using SUSENAS 2005 data, we conducted study about poverty hotspot detection using flexible scan statistics in FlexScan 3.05 as the alternatives to handle arbitrary shape of cluster and low power of test in circular spatial scan statistics. The result of the study has showed that flexible scan statistics showed good performance in detecting poverty hotspot, either first, second or third priority area. The average of reliabilty score is 91.15%. Among seven maximum cluster size used in this study, maximum cluster size 10% showed lowest average reliability and the highest average of reliability was shown when maximum cluster size were 20% and 30%.