Maximum likelihood estimation with nonlinear regression in polarographic and potentiometric studies

Abstract

In this work we review some aspects of maximum likelihood nonlinear modeling in polarographic and potentiometric techniques. Different algorithms, namely the Levenberg–Marquardt and the “error-in-variables” methods in parametric and Monte-Carlo nonparametric estimation are used. Conclusions are drawn upon the influence of experimental errors and error correlation, introduced via statistical weighting, in the accuracy and precision of the estimated parameters. Several of the tested alternatives, including regression on the signal variable alone with a global error weighting function, are shown to provide adequate representation of the experimental data.