fitting exponential decay with no initial guessing

Question

Does anyone know a scipy/numpy module which will allow to fit exponential decay to data?

Google search returned a few blog posts, for example - http://exnumerus.blo

Answer 1

Procedure to fit exponential with no initial guessing not iterative process :

This comes from the paper (pp.16-17) : https://fr.scribd.com/doc/14674814/Regressions-et-equations-integrales

If necessary, this can be used to initialise a non-linear regression calculus in order chose a specific criteria of optimisation.

EXAMPLE :

The example given by Joe Kington is interesting. Unfortunately the Data isn't shown, only the graph. So, the data (x,y) below comes from a graphical scan of the graph and as a consequence the numerical values are probably not exactly those used by Joe Kington. Nevertheless, the respective equations of the "fitted" curves are very close one to the other, considering the wide scatter of the points.

The upper Figure is the copy of the Kington's graph.

The lower Figure shows the results obtained with the procedure presented above.