COLLECTIVE OF NONPARAMETRIC REGRESSIONS BASED ON DECOMPOSITION OF TRAINING SAMPLE

The technique of synthesis and the analysis of collective nonparametric regresses providing high computing efficiency of the decision of problems of restoration of stochastic dependences due to use of technology of parallel calculations is offered. Asymptotic properties of collective are researched, its comparison with traditional nonparametric regression is made.

nonparametric regression, the big samples, asymptotic properties, principles of decomposition, collective estimations, parallel computing technologies