Bayesian networks with continuous and/or discrete variables can be learned and compared from data. The method is described in Boettcher and Dethlefsen (2003), <doi:10.18637/jss.v008.i20>.
