How should I be applying a Poisson fit to this dataset?

Question

I followed this document trying to fit a set of frequency data to Poisson distribution in R, but goodfit throws many warnings:

>library(vcd)
>freq
[1]  71   9   1   7  27 158   1   6   1  28   1   7  33  10   2  14   3   5  17  73   8   248
[23]   6 129  11  15
>gf<-goodfit(freq,type="poisson",method="MinChisq")
Warning messages:
1: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
2: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
3: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
4: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
5: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
6: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
7: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
8: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value
9: In optimize(chi2, range(count)) :
  NA/Inf replaced by maximum positive value

freq is an integer vector. It's taken from a larger integer vector to illustrate the problem here. The larger integer vector have a distribution similar to Poisson.

What's wrong with freq, or is it a bug of the package vcd?

I can't find object `freq` in the pdf you link to. Am I missing something? Is `freq` your object? If so, I think we'll need a reproducible example. — Roman Luštrik, Mar 11 '14 at 07:15
The way it is phrased confused me: it seems like you want to replace NA/Inf with the maximum value. — PascalVKooten, Mar 11 '14 at 08:24
Have you plotted your dataset to see if it looks at all like a Poisson distribution? — Carl Witthoft, Mar 11 '14 at 11:53
Sorry for the confusion, the "NA/Inf replaced by maximum positive value" part is the warning thrown by `goodfit` — Zebra Propulsion Lab, Mar 11 '14 at 16:46

How should I be applying a Poisson fit to this dataset?

0 Answers0