A novel Bayesian approach for latent variable modeling from mixed data with missing values

We consider the problem of learning parameters of latent variable models from mixed (continuous and ordinal) data with missing values. We propose a novel Bayesian Gaussian copula factor (BGCF) approach that is proven to be consistent when the data are missing completely at random (MCAR) and that is...

Full description

Saved in:
Bibliographic Details
Published inStatistics and computing Vol. 29; no. 5; pp. 977 - 993
Main Authors Cui, Ruifei, Bucur, Ioan Gabriel, Groot, Perry, Heskes, Tom
Format Journal Article
LanguageEnglish
Published New York Springer US 11.09.2019
Springer Nature B.V
Subjects
Artificial Intelligence
Bayesian analysis
Computer simulation
Continuity (mathematics)
Mathematics and Statistics
Probability and Statistics in Computer Science
Statistical Theory and Methods
Statistics
Statistics and Computing/Statistics Programs
Latent variables
Gaussian copula factor model
Parameter learning
Missing values
Mixed data
Online AccessGet full text

Cover

More Information
Summary:We consider the problem of learning parameters of latent variable models from mixed (continuous and ordinal) data with missing values. We propose a novel Bayesian Gaussian copula factor (BGCF) approach that is proven to be consistent when the data are missing completely at random (MCAR) and that is empirically quite robust when the data are missing at random, a less restrictive assumption than MCAR. In simulations, BGCF substantially outperforms two state-of-the-art alternative approaches. An illustration on the ‘Holzinger & Swineford 1939’ dataset indicates that BGCF is favorable over the so-called robust maximum likelihood.
ISSN:0960-3174
1573-1375
DOI:10.1007/s11222-018-09849-7