Question

I have a dataset with zipf law distribution, and now i want compare it with standard zipf diagram, this is my R code:

f <- read.table('/myfile.txt',sep='\t',header=T)
attach(f)
fs <- f[order(cnt), ]
detach(f)
n = 1:dim(fs)[1]
plot(fs[,2]~n)

Now i want to compare it with zipf in the same plot, How can i do this in R?

Answer 1

The VGAM package has facilities for estimating the exponent in the Zipf distribution. You might want to plot your distribution against a best fit estimated density:

plot( cntdens <- table(f[['cnt']])/length(f[['cnt']]),
       xlim=range(f[['cnt']]), ylim=c(0, 0.8)  )

# And then plot the theoretic distribution for the VGAM fit 
# ... extending the example on ?VGAM::zipf:

require(VGAM)
zdata <- data.frame( y=1:max( f[['cnt']] ),  ofreq= table( f[['cnt']] ) )
fit = vglm(y ~ 1, zipf, zdata, trace = TRUE, weight = ofreq, crit = "coef")
proby = dzipf(1:max(f[['cnt']]), N =max(f[['cnt']]), s = Coef(fit) )
points((1:5)+0.05, proby,  col="red")

Below is that procedure using the data in the example on the ?dzipf page.