The determination of a differential equation underlying a measured time ser
ies is a frequently arising task in nonlinear time series analysis. In the
validation of a proposed model one often faces the dilemma that it is hard
to decide whether possible discrepancies between the time series and model
output are caused by an inappropriate model or by bad estimates of paramete
rs in a correct type of model, or both. We propose a combination of paramet
ric modelling based on Beck's multiple shooting algorithm and nonparametric
modelling based on optimal transformations as a strategy to test proposed
models and if rejected suggest and test new ones. We exemplify this strateg
y on an experimental time series from a chaotic circuit where we obtain an
extremely accurate reconstruction of the observed attractor. (C) 2000 Elsev
ier Science B.V. All rights reserved.