SAMPLE TESTING OF THE ALGORITHM WITH THE ITERATIVE DETERMINATION OF WEIGHTS

Abstract

The article presents the results of testing the algorithm robust estimation based on the Chebyshev inequality, on a large number of samples with unimodal symmetric distribution with an asymmetric noise. Difference of the considered way of estimation, the using weight of supervision, from different ways of weighing of supervision, is that function of weighing is based by analogy with Chebyshev's inequality which doesn't depend on distribution type. It means that for the used way of estimation it isn't required aprioristic knowledge like distribution, the method is free from distributions. The main result indicates that the iterative algorithm using a first approximation to the parameter of the window function suitable values of both conventional (unstable) standard deviation, and value means the accuracy of the measuring tool. Keywords: robust estimation, the Chebyshev inequality, testing, weighing function

Keywords

• robust estimation
• the chebyshev inequality
• testing
• weighing function