EcoSta 2022: Start Registration
View Submission - EcoSta2022
A0776
Title: Asymptotic behaviors of hierarchical clustering under high dimensional settings Authors:  Kento Egashira - University of Tsukuba (Japan) [presenting]
Kazuyoshi Yata - University of Tsukuba (Japan)
Makoto Aoshima - University of Tsukuba (Japan)
Abstract: Hierarchical clustering has been approved as a useful tool for the analysis of gene expression microarray data on behalf of high-dimensional, low-sample-size (HDLSS) data. While three asymptotic behaviors of hierarchical clustering are deliberated under asymptotic settings from moderate dimension through HDLSS, it is considered that the conditions required are strict for HDLSS data due to having discussions under several asymptotic settings at once. Given this background, this presentation focuses on HDLSS settings and we prove the asymptotic properties of hierarchical clustering under mild and practical settings for HDLSS data. We proceed with the current comprehension of hierarchical clustering under HDLSS settings without assuming normality. Finally, numerical simulation studies are given and we discuss the performance of the hierarchical clustering for high dimensional data.