Analysing curves using kernel estimators

Publisher Website
Google Scholar
Abstract
In this paper a novel statistical method for curve fitting is described and applied to growth data for illustration. This technique, called kernel estimation, is non-parametric and belongs to the class of smoothing methods. Therefore, it does not need an a priori functional model where individual parameters are determined from the data. Functional models can only reflect features which have been incorporated into the model. Recent progress in selecting the degree of smoothing from the data makes the new method more easy to use and more objective. It applies to the curve itself or to its derivatives.

This publication has 11 references indexed in Scilit: