Re: [igraph] fitting a straight line to power-law distribution

[Fabio Daolio]

Hi Narcis,

that is because, if I got it right, when a probability density function p(x) is 
power-law
with an exponent alpha, i.e. p(x)~x^(-alpha),
then its complementary cumulative distribution function is also power-law (just 
do
the integration from x to infinity) but with exponent -alpha+1, i.e. 
P(x)~x^(-alpha+1)

On that note, I've been pointed out to a good paper about fitting empirical data
http://arxiv.org/abs/0706.1062

Sincerely,
--
Fabio

On Jul 28, 2011, at 11:25 AM, Narges Zarrabi wrote:

> Hi,
> 
> I am trying to understand the fitting to the power law distribution in the 
> example in link below:
> http://igraph.sourceforge.net/screenshots2.html#8
> 
> However, in the following line of the example:
> lines(10:500, 10*(10:500)^(-coef(alpha)+1))
> 
> I don't understand why +1 is added in the power (-coef(alpha)+1)
> 
> Could anyone explain the reason ? Thanks
> 
> Regards,
> Narcis
> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help