Supports Bayesian models with full and partial (hence
arbitrary) dependencies between random variables. Discrete and continuous
variables are supported, and conditional joint probabilities and probability
densities are estimated using Kernel Density Estimation (KDE). The full
general form, which implements an extension to Bayes' theorem, as well as
the simple form, which is just a Bayesian network, both support regression
through segmentation and KDE and estimation of probability or relative
likelihood of discrete or continuous target random variables. This package
also provides true statistical distance measures based on Bayesian models.
Furthermore, these measures can be facilitated on neighborhood searches,
and to estimate the similarity and distance between data points.
Related work is by Bayes (1763)