A Bayesian Approach to Spherical Factor Analysis for Binary Data

39 Pages Posted: 28 Sep 2020

See all articles by Xingchen Yu

Xingchen Yu

University of California, Santa Cruz

Abel Rodriguez

University of California, Santa Cruz

Date Written: August 11, 2020

Abstract

Factor models are widely used across diverse areas of application for purposes that include dimensionality reduction, covariance estimation, and feature engineering. Traditional factor models can be seen as an instance of linear embedding methods that project multivariate observations onto a lower dimensional Euclidean latent space. This paper discusses a new class of geometric embedding models for multivariate binary data in which the embedding space correspond to a spherical manifold, with potentially unknown dimension. The resulting models include traditional factor models as a special case, but provide additional flexibility. Furthermore, unlike other techniques for geometric embedding, the models are easy to interpret, and the uncertainty associated with the latent features can be properly quantified. These advantages are illustrated using both simulation studies and real data on voting records from the U.S. Senate.

Keywords: Spherical factor model, Dimensionality reduction, Riemannian manifold, Bayesian methods

Suggested Citation

Yu, Xingchen and Rodriguez, Abel, A Bayesian Approach to Spherical Factor Analysis for Binary Data (August 11, 2020). Available at SSRN: https://ssrn.com/abstract=3672055 or http://dx.doi.org/10.2139/ssrn.3672055

Xingchen Yu (Contact Author)

University of California, Santa Cruz ( email )

1156 High St
Santa Cruz, CA 95064
United States

Abel Rodriguez

University of California, Santa Cruz ( email )

1156 High St
Santa Cruz, CA 95064
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
37
Abstract Views
227
PlumX Metrics