ABSTRACT
The Bayesian Factor Analysis model is similar to the Bayesian Source Separation model in that the factors, analogous to sources, are unobservable. There are two main reasons why one would perform a Factor Analysis. The first is to explain the observed relationship among a set of observed variables in terms of a smaller number of unobserved variables or latent factors which underlie the observations. The second reason one would carry out a Factor Analysis is for data reduction. Since the observed variables are represented in terms of a smaller number of unobserved or latent variables, the number of variables in the analysis is reduced, and so are the storage requirements. The structure of the Factor Analysis model is strikingly similar to the Source Separation model. The main results of performing a Factor Analysis are estimates of the factor score matrix, the factor loading matrix, and the error covariance matrix.
