Bayesian maximal likelihood estimation for multidimensional combinatorial predictions

multi-dimensional combinatorial forecasting;  Bayesian maximum likelihood estimation;  Concordant Differential Array

In order to meet the needs of actual forecasting, the multidimensional combination prediction problem is proposed, that is, the variable weight combination prediction of several single multidimensional predictions. Under the premise that each single prediction is unbiased and obeys the normal distribution, the prior distribution density and posterior distribution density of the p moment prediction vector Xp are found. Using subjective prior information, prediction information and sample information, and Bayesian estimation method, Bayesian maximum likelihood estimation of Xp is obtained, and its weight changes with the change of p. This method makes full use of the relevant information between multidimensional variables, further improves the scientificity and effectiveness of prediction, and reflects the further comprehensive application of sample information and prediction information.