Home/Working Papers/Parsimonious Bayesian Factor Analysis when the Number of Factors is Unknown
Parsimonious Bayesian Factor Analysis when the Number of Factors is Unknown
Código: WPE – 345
Hedibert Freitas Lopes
We introduce a new and general set of identifiability conditions for factor models whichhandles the ordering problem associated with current common practice. In addition, thenew class of parsimonious Bayesian factor analysis leads to a factor loading matrix representationwhich is an intuitive and easy to implement factor selection scheme. We argue thatthe structuring the factor loadings matrix is in concordance with recent trends in appliedfactor analysis. Our MCMC scheme for posterior inference makes several improvementsover the existing alternatives while outlining various strategies for conditional posterior inferencein a factor selection scenario. Four applications, two based on synthetic data andtwo based on well known real data, are introduced to illustrate the applicability and generalityof the new class of parsimonious factor models, as well as to highlight features of theproposed sampling schemes.