Description: In the article the model of the educational process as a Markov some approximation with discrete states and continuous time is considered. All transitions from state to state, occurring under the influence of some streams of events, presented in the form of a mark-up of the state graph. Such graph is a "scheme of birth and death". The question of the terminology used, analyzes the characteristics of the design of Bayesian classifiers under parametric a priori uncertainty is considered. The practical application of the research object in the diagnostics of the quality of the educational process at substantially limited amounts of training sample is considered.
Keywords: quality of the educational process, process of «birth and death», priori uncertainty, limited sample, Bayesian classifier, volume of training sample, probability of error of classifier