observed and predicted plot of count data

[Madu Abuchi]

Any help please?

I am after a stata code to help plot the observed and predicted count of data following comparison with Poisson and negative binomial.

After some search, I found this stata user written command -prcounts- . However, it does generate the predicted estimates but does not plot the graph. I am not sure if my Stata/IC 15 is supported as it seems the code was written since the year 2000 http://www.indiana.edu/~jslsoc/files...TBprcounts.pdf .

Sample of my code is:

* Example generated by -dataex-. To install: ssc install dataex
clear
input int pre float(counts disease)
 3  1 2
 3  2 2
 3  3 2
 3  4 2
11  1 2
 3  1 2
 3  2 4
11  1 2
11  2 4
32  1 2
 3  1 2
 3  2 2
 0  0 2
 3  1 2
 3  1 2
 3  2 2
 3  3 2
 3  1 2
 3  2 2
 0  0 2
 3  1 2
 3  2 2
 3  3 2
 3  1 2
 3  2 4
 3  3 2
 3  1 2
 0  0 1
 3  1 2
 7  1 2
 3  1 2
 3  2 2
 3  1 2
 3  2 2
 3  3 2
11  1 2
11  2 2
 0  0 2
 1  1 2
 3  1 4
 3  2 2
 3  3 2
 3  4 2
 3  5 2
 3  6 1
 3  7 2
 3  8 2
 3  9 2
 3 10 2
 3 11 1
 3 12 2
 3 13 2
 3 14 2
 3 15 2
 3 16 2
 3 17 1
 5  1 2
11  1 2
11  2 4
11  3 4
11  4 2
18  1 2
32  1 2
 0  0 1
 3  1 2
 3  2 2
 3  3 2
 3  4 2
 3  5 2
 3  6 2
 3  7 2
 3  8 2
11  1 2
 0  0 1
 3  1 2
 3  2 2
 3  3 2
 3  4 2
 3  5 2
 3  6 2
 3  7 2
 3  8 2
 3  9 2
 3 10 2
 3 11 2
 3 12 2
 3 13 2
11  1 4
11  2 4
11  3 2
 3  1 2
 3  2 2
 3  3 2
 3  4 2
 3  5 2
 3  6 2
11  1 2
 2  1 2
 2  2 2
 3  1 2
end

poisson counts i.disease
prcounts pois, max(16) plot

nbreg counts i.disease
prcounts nbreg, max(16) plot

graph poisrate poisval nbregrate nbregval ziprate

I get the following message if I try to graph:

poisrategraph_g.new poisval nbregrate nbregval ziprate: class member function not found
r(4023);

All I want is a graph like the one attached:

Thanks

[Nick Cox]

ziprate is not in your example data. Aother problem is that your graph syntax was superseded by Stata 8

 
 version 7: graph poisrate poisval nbregrate nbregval ziprate

may work (i.e. be a legal command) with your data, but it's not the graph you ask for. Further, you are fitting with a categorical predictor, so your set-up is one distribution for each category and each model, so you need something more complicated.

[Madu Abuchi]

Hi Nick,

Thanks, I have used the version control command as suggested

 
 version 7: graph poisrate poisval nbregrate nbregval

But that is not what I am expecting. Is there any command you are aware of that I can use to get the graph I'm expecting.

Thanks

[Nick Cox]

I don't know what graph you are expecting, so you should explain. I count 3 diseases in your example data. Perhaps there are even more in your full data.

[Madu Abuchi]

My aim is to compare Poisson and negative binomial regression in this example count data. My predictor variable (disease) in full data is 0 to 4. I want to know which of the method fits my data better than each other graphically. In my full data I'm extending to include zero-inflated Poisson and zero inflated negative binomial

[Nick Cox]

Sure, I understand that but I am not clearer on what graphs you expect.

[Madu Abuchi]

Something like this

[Giorgia Estefani]

Hi Madu, I am wondering if you found a solution to your problem above. I have the same. Many thanks!

[Nick Cox]

Previous posts got nowhere because each model fitted in #1 implies a set of distribution curves, not one.

[Giorgia Estefani]

thanks, nick, for your reply. but there is any way in Stata to plot (together with the histogram graph of the observed distribution) a line depicting with a binomial distribution with a p (the average probability of the observed distribution) and an n? and maybe a test that says how much the two look like the same? sorry for silly questions. many thanks

[Nick Cox]

So you don't have the same problem; you have a different problem. In any case please do read https://www.statalist.org/forums/help#stata and give some example data. How many trials?