Loading Events

Generalizable Stability Analysis for Clustering Algorithms by Eric Rosenthal ’19

Wed, April 24th, 2019
1:10 pm
- 1:50 pm

Image of Stetson Court classroom

Generalizable Stability Analysis for Clustering Algorithms by Eric Rosenthal ’19, Wednesday, April 24, 1:10 – 1:50 pm, Statistics Senior Thesis Defense

Abstract:  We introduce a novel stability based measure of cluster validity designed to be adaptable for density based clustering and spatiotemporal clustering. We examine the performance of this stability measure on the DBSCAN and ST-DBSCAN clustering algorithms to demonstrate its effectiveness at finding stable clusterings, and use it to examine clusterings of weather data from Europe and voter registration from California. We also discuss methods of simulating the sampling distribution of spatiotemporal data to be used with our measure.

Event/Announcement Navigation

Related Events