scientific article; zbMATH DE number 3884710

From MaRDI portal
Publication:3218586





[For the entire collection see Zbl 0533.00014.] Adaptive approximation techniques are needed when approximating discrete sets of points. Most useful approximating sets of functions are splines, and their multivariate extensions, also exponential families. The question, which is considered here, is, what criteria can be used to determine how smooth we need to approximate the data so that the residuals of the approximation are the errors in the data values? Three methods are presented. The first comparing the unit-lag auto-correlation of the sequence of residuals for subsequent approximations. The second utilizes the generalized cross-validation technique, while the third uses information theoretic criteria to arrive at the Akaike information criterion.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3218586)