[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: [igraph] Question about fitting a line to a power-law distribution

**From**: |
Gábor Csárdi |

**Subject**: |
Re: [igraph] Question about fitting a line to a power-law distribution |

**Date**: |
Mon, 14 Feb 2011 10:57:23 +0100 |

On Mon, Feb 14, 2011 at 12:49 AM, Narges Zarabi <address@hidden> wrote:
>* Dear all,*
>* I am trying to fit a straight line to the degree distribution of my network*
>* (which is scale-free), using following commands:*
>* g <- barabasi.game(1000, directed =F)*
>* d <- degree(g)*
>* dd <- degree.distribution(g,cumulative=TRUE)*
>* plot(dd,log="xy")*
>* alpha <- power.law.fit(d, xmin=10)*
>* lines(5:50, 10*(5:50)^(-coef(alpha)+1))*
>* but if I change the value xmin, the value of alpha changes and hence*
>* different straight line are generated! Is there a way to know which xmin is*
>* best to use? and which alpha and fitted line is more accurate?*
Dear Narges,
I would fix an xmin value (say 20) and generate much larger networks,
e.g. with 1 million nodes, to have enough high degree vertices for the
fit. But I'm afraid there is no really principled way to determine
xmin, at least I don't know any. The larger your network is the less
it matters, AFAIK.
Best,
Gabor
>* Thanks in advance,*
>* Narsis*
>* _______________________________________________*
>* igraph-help mailing list*
>* address@hidden*
>* http://lists.nongnu.org/mailman/listinfo/igraph-help*
>
>
--
Gabor Csardi <address@hidden> UNIL DGM