Restoration of statistical dependences by averaged splines

The use of polynomial splines when there is a deficiency of a priori information about the dependence to be restored is considered. Conditions are found under which the splines solving the problem of least squares (averaged splines) uniformly approximate the regression function and its derivatives.