Heterogeneous Data Analysis Based on HDLSS Asymptotics

• Activity Number: 287
• Type: Invited
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): James Stephen Marron*
• Companies: The University of North Carolina
• Keywords: big data ; Gaussian mixture ; High Dimension

Abstract

A major statistical challenge in the age of Big Data is data heterogeneity, which is endemic to the combination of multiple data sets. While statistical thinking on this issue is in infancy, some ideas, and potential approaches are discussed. A useful thought model is Gaussian mixtures, however it is important to realize that classical mixture estimation is impossible, in most big data situations where the dimension is very large. Mathematic statistics are explored using the fairly new concept of High Dimension Low Sample Size asymptotics.