Modelos não lineares e suas aplicações

Abstract

The main objective in this work is to present a review of the normal nonlinear mo- dels, contemplating their classes, the least squares method to estimate its parameters,

emphasizing the iterative method of Gauss-Newton and Newton-Raphson. Still, some asymptotic results, measures of nonlinearity and diagnostic techniques are presented. A characteristic of the nonlinear models is that diagnostic techniques used in linear models are inadequate if the degree of nonlinearity is high. In this work, we studied the existing nonlinearity measures for the verification of these assumptions and we use them in our

applications. Finally, we analyze the data set European Rabbits to demonstrate the ap- plicability of a nonlinear model proposed in the literature. Also, to the same data set we adjust other nonlinear models, including estimation methods for nonparametric curves, for comparison purposes. We conclude that in nonlinear models there is a strong link between the nonlinear functional and the phenomenon under study.