A dimensionality reduction process to forecast events through stochastic models

Abstract : This paper describes a dimensionality reduction process to forecast time series events using stochastic models. As well as the KDD process defines a sequence of common steps to achieve useful information through data mining techniques, we propose a sequence of steps in order to estimate the probability of future events through stochastic modeling. Our process focus on reduce the dimension-ality of data, thus reducing the effect of the common problems involved in stochastic modeling, such as the state space explosion and the large modeling efforts to create such models.