Probabilistic data based modelling: a new technique for data analysis and process modelling

A probabilistic data based modelling algorithm is introduced. This frequentist algorithm for data based modelling is very different from the more commonly used techniques like neural networks and NARMAX modelling and which attempt to construct a functional representation of a reference data set. The probabilistic algorithm has several advantages over these techniques. It calculates predictions based upon an estimate of the multi-dimensional data distribution, allowing the construction of meaningful confidence bounds and other probability measures useful for model analysis. Two statistics used for the analysis of dynamic data are introduced, and their use with a probabilistic modelling algorithm is illustrated with examples.