A measure of nonlinearity in singular nonlinear regression models

摘要

The method used in nonlinear regression is most of the time the nonlinear squares. If the nonlinearity of the model is not too strong, it is possible to linearize the model and to use the linear statistical theory which is more simple. The question is under which condition this can be done. If the linearized model is regular such criteria were found. To find a solution for singular models is the aim of the paper.