Abstract: A method for predicting an optimal mixture of power generation plant types using a forecasting model, the method includes: collecting financial data of estimated costs of each of a plurality of power generation plant types; assigning a probability distribution to a plurality of the estimated costs; projecting the estimated costs over a period of time for each type of power generation plant to generate probability distributions of estimated costs for each type of power plant; applying a Bayesian combination analysis to evaluate non-quantifiable factors that influence the probability distributions, and projecting a mixture of power plant type usage based on the Bayesian combination analysis.