Hi all:
I have 3 questions about the poisson regression of contingency table.
Q1¡¢How to understand the "independent poisson process"as many books or paper
mentioned?
For instance:
Table1
-------------------------------------------
treat caner non-cancer sum
-------------------------------------------
treat1 52(57.18) 19(13.82) 71
treat2 39(33.82) 3 (8.18) 42
-------------------------------------------
sum 91 22 113
-------------------------------------------
The numbers in parentheses are the expected frequency,which are computed
by(57.18 for instance):
71*91/113=57.18
But according to poisson distribution,the distribution function is:
pr<-exp(-mu)*mu^x/x!.But I can't find this function in the computation of
expected frequency of Table1.
In other words,how the underdstand the"independent poisson process" from
Table1?
Q2¡¢How to understand "zero-inflated"?
For instance:
area nation frequency
1 1 10
1 2 0
1 3 0
1 4 0
2 1 20
2 2 0
2 3 0
2 4 5
3 1 15
3 2 6
3 3 20
3 4 0
Since most of the frequencies(6/12) are zero,so I can simply consider the
dataset is "zero-inflated"?
In other words, can I judge the dataset is "zero-inflated" simply by the
proportion of "zero frequency"?
Or,how to judge whether the dataset is "zero-inflated" or not?
Q3¡¢How to understand "over-dispersion"?
For instance:
area nation frequency
1 1 10
1 2 0
1 3 0
1 4 0
2 1 20
2 2 0
2 3 0
2 4 5
3 1 15
3 2 6
3 3 20
3 4 0
Obviously, the frequencies' distribution is of great skewness.So can I judge
the dataset is "over-dispersion"?
In other words, can I judge the dataset is "over-dispersion" simply by the
distribution of all the frequencies?
Or,how to judge whether the dataset is "over-dispersion" or not?
Thanks a lot for your help.
My best.
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
