[R] power law fit with unknown zero

[Philipp Pagel]

> I would like to do a fit of the form: y = a (x+c)**b, where a, b and c 
> are unknown.

The popular method of fitting a linear model to log-log transformed data was
already mentioned. ALthough this works in many situations, it is
considered to be somewhat biased an not very robust. I suggest to have a
look at the following papers for some good suggestions:

Goldstein, M.; Morris, S. & Yen, G. 
Problems with Fitting to the Power-Law Distribution 
European Physical Journal B, 2004, 41, 255-258

Newman, M. E. J. 
Power laws, Pareto distributions and Zipf's law
Contemporary Physics, 2005, 46, 323

cu
	Philipp


-- 
Dr. Philipp Pagel                              Tel.  +49-8161-71 2131
Lehrstuhl für Genomorientierte Bioinformatik   Fax.  +49-8161-71 2186
Technische Universität München
Wissenschaftszentrum Weihenstephan
85350 Freising, Germany

 and

Institut für Bioinformatik / MIPS
GSF - Forschungszentrum für Umwelt und Gesundheit
Ingolstädter Landstrasse 1
85764 Neuherberg, Germany
http://mips.gsf.de/staff/pagel