ON ESTIMATION OF PARAMETERS OF BETA DISTRIBUTIONS BY GROUPED OBSERVATIONS WITH CENSORING

In this work we study the problem of parameters estimation of the Beta distribution with a specific observation pattern. The observations are assumed to group and full non-observations of individual realization values are possible. Application of a modified in a certain sense, method of maximum likelihood is suggested and the offered procedure is shown to lead to consistent estimators. The experiment is designed in such a way that the actual number of realizations is unknown to us, we know only a part of those realizations.