East Coast Ballroom
Simultaneous Ranking and Clustering of Small Areas based on Health Outcomes using Nonparametric Empirical Bayes Methods (306666)
Cora Allen-Coleman, University of Wisconsin-Madison*Ronald Gangnon, University of Wisconsin-Madison
Keywords: ranking, clustering, small area estimation, empirical Bayes
A common task is ranking different geographic units (small area), e.g. counties in the United States, based on health (or socioeconomic) outcomes/determinants. We propose a nonparametric empirical Bayes (finite mixture) model for small area health outcomes that is suitable for simultaneous ranking and clustering of small areas. Optimal point estimates of the ranks are obtained to minimize expected integrated squared error loss on the health outcome (mean or proportion) scale. Small areas are simultaneously clustered by assigning each small area to the optimal (minimum mean square error loss) cluster (mixture component) for its estimated rank position. We illustrate our approach using an analysis of percent low birth weight births for Wisconsin counties, 2008-2014.