Judging a Fit I: Real x
Fitting a function β to the observed x may yield a poor result even if the available parameters are ξ optimised As explained at the beginning of Chap. 13, the function β(t) is actually a family of functions β(t; ξ), and one determines the range of ξ that falls into the Bayesian area. It is possible that even the optimum parameter ξ opt, defined in Sect. 3.3, yields an inadequate fit. The optimum parameter suggests that one can represent the observed data by the function β pre = β(t; ξ opt).If β pre is not adequate, no member of the family β reasonably fits x.
Unable to display preview. Download preview PDF.