R help - Dec 2008 - Zipf fitting using R

Dear R-users,

 

I am new to R and would like to use it for fitting the zipf distribution to
some numeric data that I have. Here's the snippet that I use:

 

library(VGAM)

X <- read.table(file("~\\mydata.txt", encoding="latin1"))

w <- as.vector(t((X[2])))

w <- w/sum(w)

y <- (1:length(w))

fit = vglm  (y ~ 1, zipf, tra=TRUE, weight=w)

zipf(N=NULL, link="loge", earg=w, init.s=NULL)

coef(fit, matrix=FALSE)

 

 

Though R returns the necessary values, I am not entirely sure if this value
returned (by the last command),is log(s) or log(-s), where s is the zipf
exponent.

 

Any clarification would be greatly appreciated! Thank you in advance!


	[[alternative HTML version deleted]]

Hi there,

I haven't seen any answers to your question on the list, so here's a  
rather late response.

I'm not familiar with the VGAM package, but if you just want to fit  
Zipf's law (did you mean Zipf's law or an actual probability  
distribution?) to some data, why not do this directly with nls() or  
optim()?  If you pass the Zipf equation to those functions, they  
should give you estimates for the parameters you're interested in.

If you're particularly interested in low-frequency data (i.e. the long  
tail of the Zipf ranking) and actual probability distributions, you  
could try our "zipfR" package available from CRAN.  The tutorial and  
background materials at

	http://zipfR.R-Forge.R-project.org/

should help you to get started and will also explain how to calculate  
the coefficients of Zipf's law from estimates of the model parameters.


Best regards,
Stefan Evert

[ stefan.evert@uos.de | http://purl.org/stefan.evert ]



On 27 Dec 2008, at 10:01, Epithemeus wrote:
> Dear R-users,
>
> I am new to R and would like to use it for fitting the zipf  
> distribution to
> some numeric data that I have. Here's the snippet that I use:
>
> library(VGAM)
>
> X <- read.table(file("~\\mydata.txt",
encoding="latin1"))
>
> w <- as.vector(t((X[2])))
>
> w <- w/sum(w)
>
> y <- (1:length(w))
>
> fit = vglm  (y ~ 1, zipf, tra=TRUE, weight=w)
>
> zipf(N=NULL, link="loge", earg=w, init.s=NULL)
>
> coef(fit, matrix=FALSE)



	[[alternative HTML version deleted]]