(NOTE: This pacakge is now incorporated into EMMIXmfa.)
Fits a mixture of factor analyzers with a common component matrix for the factor loadings before the transformation of the latent factors to be white noise. It is designed specifically for the task of displaying the observed data points in a lower (q-dimensional) space, where q is the number of factors adopted in the factor-analytic representation of the observed vector.
It also provides a greater reduction in the number of parameters in the model. Component distributions can either be from the family of multivariate normals or from the family of multivariate t-distributions. Maximum likelihood estimators of model parameters are obtained using the Expectation-Maximization algorithm.
Fitting a MCFA model with there components using two factors for the Iris data available in R can be done using,
fit <- mcfa(Y = iris[, -5], g = 3, q = 2)
The groupings can be visualized in the q-dimensional factor space.
plot_factors(fit)
MCFA fits multivariate normals to the data, fitting t-distributions can be achieved
using mctfa function. Further, there are functions to generate data from a emmix-mcfa
models (rmix), estimate factor scores (factor_scores), estimate adjusted Rand Index (ari),
find the number of misallocations (err), among others.