Algorithm and EXCEL Application for Data Stepwise Approximation by Solution of the System of Equations Using Gauss Method

Abstract

The paper presents the algorithm and the application in Excel of the stepwise least squares data approximation for a polynomial model linear with respect to coefficients. The term approximating polynomial is understood as a generalized polynomial which monomials are any linearly independent functions. The coefficients of the approximating polynomial are determined by the Gaussian method. Specially created information table enables an optimal selection of monomials and building a model step by step, by incorporating the monomials with most decreasing sums of squared deviations. In this table, the sum of squared deviations, extreme deviations and standard deviation values are indicated at each approximation step. This is the basis for making the decision to complete the approximation, as well as the selection of points with excessive deviations.