Parametric and Nonparametric Identification of the Distribution Law from Interval Data

Abstract

The properties of nonparametric estimation of the distribution function and estimations of the maximal likelihood of the parameters of distributions from interval data are considered. Samples of interval observations are required when the true value of an observation is not known, and only the interval this value belongs to is known. An investigation of the statistical properties of estimates of the distribution function and its parameters from interval data is performed by methods of statistical simulation and compared to estimates obtained from point samples of the midpoints of the observation intervals.