This work investigates the application of Bayesian methodologies to the classification and forecasting problems. It begins with reviews of basic Static Bayesian Networks (SBNs) and the methods of probabilistic inference from SBNs, including Pearl''s causal tree and Lauritzen and Speigelhalter''s NP-complete clique tree methodologies. Pattern classification utilizing the Naïve Bayes and the Tree Augmented Naïve Bayes (TAN) classifiers is then described. Boosting and other ensemble methods are applied so as to improve performance. Attention is then turned to incorporating a time element into Bayesian Networks to create a series of SBNs, acting as time-slices to construct a Dynamic Bayesian Network, to solve the forecasting problem. Each SBN, consisting of the suitably modified TAN, and computed with correlation and partial correlation coefficients, is then combined with the Pearl causal tree. As with the classification problem, boosting is applied to improve forecasting performance. This ensemble methodology would interest researchers in diverse fields as Computer Science and Finance, and those considering an alternative to, or a combination with, traditional time-series data analysis.