The application of bayesian model for estimation of the probabilities of alternatives under uncertainty, using nonnumeric, inaccurate and incomplete the expert in-formation

In this paper considered the Bayesian model of estimation of piecewise-constant density corresponding to the decomposition of the ternary range of possible values of the random quantity. The model is based on the estimation of parameters of the Dirichlet distribution for nonnumeric, inaccurate and incomplete information. The analysis is performed for the evaluation and prediction of the statistical characteristics of the CHF with respect to XDR. To compare the quality of the result for the same data investigated by using classical econometric method: construction of ARIMA – models and forecasting method of exponential smoothing.