Title: Diagnostics in finite mixture models and model-based clustering
Authors: Ranjan Maitra - Iowa State University (United States) [presenting]
Abstract: Model-based clustering offers a principled approach to the problem of partitioning data into different groups. Data quality is important as is our confidence in the acquired clustering. We investigate the use of and develop regression-style diagnostics to quantify uncertainty in the groupings of observations as well as to identify influential and outlying observations with a view to improving model-fitting and inference.