Statistical and Predictive Inference Methods for Topological Data Analysis
Jordan Schupbach Ph.D. Proposal in Statistics (Dept. of Mathematical Sciences, MSU)
06/13/2022
Abstract:
Topological data analysis (TDA) is an interdisciplinary field that seeks to represent
the shape of data using tools from algebraic topology. However, methods for analyzing
these representations under non-trivial sampling designs are few to non-existent.
As a result, statistical inference is rarely employed in practice and estimates of
uncertainty typically assume independence
when conducting predictive inference. In this talk, methods for conducting statistical
and predictive inference in TDA for hierarchical sampling designs are proposed. In
particular, collections of persistence diagrams (a common topological descriptor)
given by a hierarchical sampling design are proposed to be modeled as a replicated
mixed-effect inhomogeneous point process. With this model, we propose how predictive
and statistical inference can be conducted in TDA and we propose some solutions to
computational issues that arise in this context.