Factor Analysis is a way of describing observed (and possibly) correlated input variables in terms of lesser number of unobserved or latent variables.

It is somewhat correlated to the PCA technique, but factor analysis uses more domain-specific knowledge in order to estimate the number of latent variables.