SEQUENTIAL PROCEDURE FOR NONPARAMETRIC ESTIMATION OF STOCHASTIC DEPENDENCE
A technique is proposed for constructing a nonparametric model of a multidimensional dependence in conditions of small volumes of initial statistical data. The method is based on a sequential procedure for building a nonparametric model in the space of argument lists of the restored dependency. The asymptotic properties of the developed model are investigated and the results of computational experiments are presented.
Keywords: nonparametric regression, small samples, sequential decision-making procedure, asymptotic property
