B0427
Title: Mixture modeling of data with hierarchy
Authors: Semhar Michael - South Dakota State University (United States) [presenting]
Andrew Simpson - South Dakota State University (United States)
Christopher Saunders - South Dakota State Univerisity (United States)
Dylan Borchert - South Dakota State University (United States)
Larry Tang - University of Central Florida (United States)
Abstract: Finite mixtures are known for modelling heterogeneity in data. The Gaussian mixture model is the most used by practitioners. The common way of estimating the parameters of this model assumes that the data is sampled through a simple random sampling process. However, in some applications such as the forensic source identification problem, data has a hierarchical structure in addition to the heterogeneity that occurs at different levels. Identifying and characterizing subpopulations in a population is discussed when there are hierarchically structured data. This will be done through semi-supervised finite mixture models and by applying constraints that account for the hierarchy in the data. This is illustrated based on a simulation study using synthetic data and a classical glass dataset. In addition, the implications of the forensic source identification problem will be discussed.