Abstract
We investigate the properties of a modification of a hypothesis testing algorithm for topological data analysis by replacing the Wasserstein distance function, used to compute the joint loss, with the Prokhorov distance. We conclude that the Prokhorov distance achieves comparable performance, with a notably decreased tendency to report extreme p-values.
Document Type
Article
Class Name
Electrical and Systems Engineering Undergraduate Research
Language
English (en)
Date of Submission
12-8-2024
Recommended Citation
Kler, Eugene, "Improving Topology-Based Hypothesis Testing Algorithms Via Novel Distance Functions" (2024). Electrical and Systems Engineering Undergraduate and Graduate Research. 34.
https://openscholarship.wustl.edu/eseundergraduate_research/34
COinS