Id: arxiv/2611.25295 Url: https://arxiv.org/abs/2611.25295 Tags: math.ST, math.AT Score: 0 Title: Kernel Regression using Persistence Diagrams Domain: math.ST Source: arxiv Authors: Saksham Aggarwal, James Nichols, Katharine Turner Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-11-18 Page: https://delphi.science/r/arxiv/2611.25295 Persistence diagrams are widely used in topological data analysis to summarize topological and geometric structure in data. We study nonparametric regression where inputs are persistence diagrams and outputs are real-valued. Using kernel-based local averaging estimators defined directly over diagram space with Wasserstein distance, we establish consistency and derive bounds on the rate of convergence under suitable assumptions in the $L^2$-theory. We also present computational experiments that demonstrate strong performance of the proposed estimator, particularly in low data regimes.