Activity Number:
|
171
- New Nonparametric Methods for Correlated Data
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 30, 2018 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Nonparametric Statistics
|
Abstract #330258
|
Presentation
|
Title:
|
Spatial Clustering Using Spatio-Temporal Network Data
|
Author(s):
|
Ashwini Venkatasubramaniam* and Ludger Evers and Konstantinos Ampountolas
|
Companies:
|
University of Glasgow and University of Glasgow and University of Glasgow
|
Keywords:
|
clustering;
spatial;
network;
temporal
|
Abstract:
|
We propose a flexible Bayesian nonparametric clustering model for spatio-temporal data recorded using a network of sensors to identify spatially contiguous clusters. This method employs a distance dependent Chinese restaurant process (ddCRP) to accommodate the geographical constraints imposed by the network and account for the presence of limited connectivity between components of the undirected graph. We modify the non-sequential ddCRP to reduce the number of singletons by controlling the number of redundant and self-links and also utilise a Metropolis within Gibbs sampler to fully explore the space of potential cluster structures in the presence of spatial and network dependencies. This model is motivated by an application to a grid-style urban road network in downtown San Francisco, where a unique observation is assumed to be present for every space and time combination and we illustrate this model with an application to both real and simulated data.
|
Authors who are presenting talks have a * after their name.