Multidimensional weighting algorithm: features of the implementation

The paper describes implementation of an algorithm for multidimensional weighting of observations with the use of the original weighting function based on the Chebyshev inequality and the iterative approach. The results of weighing the observations are intended to obtain stable estimates in multidimensional methods (correlations, regressions, etc.); the examples of such estimates calculation are given. During iterations, the above algorithm uses renormalization over one-dimensional subspaces, i.e., it is independent of the measurement scale chosen for the individual axes of the initial multidimensional data.