Parametric and nonparametric curve fitting (Q880378): Difference between revisions

This article summarizes the convergence issues for the classical curve fitting problem. The properties of the input signal that ensure convergence of the estimate are investigated. In the parametric case, the standard least squares method is used and the sufficient conditions under which the estimated parameters converge to their true values almost surely are given. In the nonparametric case, the nearest-neighbour averaging algorithm is used and the necessary and sufficient conditions under which the estimated function almost surely converges to the true nonlinearity are shown.