Online Program Home
My Program

Abstract Details

Activity Number: 260 - SPEED: Topics in Bayesian Analysis
Type: Contributed
Date/Time: Monday, July 30, 2018 : 3:05 PM to 3:50 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #332947
Title: Uncertainty in Probabilistic Weighted Multidimensional Scaling
Author(s): Lata Kodali* and Leanna House
Companies: and Virginia Tech
Keywords: Bayesian; Multidimensional Scaling; Visual Analytics; Uncertainty

Meaningful visualizations of big data are crucial and can be misleading without including measures of uncertainty. We extend a method called Bayesian Visual Analytics (BaVA) to incorporate uncertainty as users explore data visually. BaVA relies on linear projections of data based on Weighted Multidimensional Scaling (WMDS) (Kruskal 1964). The direction in which data are projected in WMDS is determined by weights assigned to each variable. Visualizations are interactive in the BaVA framework, and possible interactions include manipulating variable weights and/or the coordinates of the two-dimensional projection (Endert et al. 2011; Leman et al. 2013). Uncertainty exists in these visualizations on the variable weights, the user interactions, and the WMDS projection. We aim to quantify these uncertainties using a probabilistic approach of WMDS under the BaVA framework (pWMDS) in order to aid users in exploration. We use posterior parameter estimates from pWMDS to assess model fit and uncertainty in the coordinates displayed. Lastly, we find an approximation to our pWMDS so that we reduce computation and enable real time data explorations that account for uncertainty.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program