Composite Indicators Building Based on Concordant of Expert-Statistical Information Using Biased Ridge Kernel Regression

The problem of constructing a complex composite indicator based on expert scores and statistical information about particular indicators is considered. A procedure for estimating the parameters of a nonlinear model of a composite indicator based on the method of biased kernel ridge regression is proposed; at the same time, the problem of optimal concordant of expert and statistical information is solved by choosing the appropriate regularization functional. Using the method of Lagrange multipliers, expressions for estimating the parameters of a nonlinear model and a kernel-based model of a composite indicator are obtained. The results of a numerical experiment on the construction of a composite indicator model using real data are presented.