Cluster-Weighted Modeling
It allows for data statistics and uncertainties to be derived from the model.
allows for well known results from past praxis to be reused in a globally non-linear context, e.g. linear system’s theory or known synthesis techniques.
Local transparency compared to conventional Neural Networks.
Fast Convergence with reasonable estimation/prediction after the first iteration.
Non-linear function approximation through soft non-linear weighting of local generalized linear models.