topic Re: Can JMP perform hypothesis test for gamma poisson distribution ? in Discussions

Can JMP perform hypothesis test for gamma poisson distribution ?

Hans_Hsu — Fri, 20 Sep 2019 09:43:03 GMT

I got data which fits gamma poisson distribution, and I'm not sure how to do hypothesis test on this kind of data.

I looked online and only found method to do hypothesis test for poisson distribution, but did not see any article for gamma poisson.

Re: Can JMP perform hypothesis test for gamma poisson distribution ?

P_Bartell — Fri, 20 Sep 2019 12:13:10 GMT

You'll need to also define the specific type of hypothesis test(s) of interest. There are many. At the very least JMP supports a variety of nonparametric hypothesis test methods.

Re: Can JMP perform hypothesis test for gamma poisson distribution ?

Hans_Hsu — Thu, 26 Sep 2019 05:58:35 GMT

I can find Hypothesis test for poisson distribution on the internet as below example, but seems JMP does not have such test (I can't find cumulative tables for poisson distribution in JMP)? And another question is when the distribution is gamma poisson, how do i modify test method?

Example:

An existing car model is known to break down on average 1.5 times/ per year. Now we want to test if new designed model is less likely to break down with ten randomly selected cars break down a total of 8 times within the first year.

Let X be the number of break downs of the new model of car in a year. X∼Poisson(λ) with λ=1.5. The null and alternative hypotheses will be

H0: λ=1.5

H1: λ <1.5.

We need to decide whether P[X≤8]<α, where α=0.05. Firstly, the expected number of breakdowns λ t =1.5×10=15. We use the cumulative tables with λ t =15 and x=8 to see P[X≤8]=0.0374

P[X≤8] = 0.0374 < 0.05

so we accept the alternative hypothesis. The average rate of breakdowns has decreased.

Re: Can JMP perform hypothesis test for gamma poisson distribution ?

Mark_Bailey — Thu, 26 Sep 2019 13:51:48 GMT

I understand that there is no direct test of the gamma poisson population parameter. Here are two things, though, that might help you.

First, you can use the JSL probability functions (PDF, CDF, and inverse CDF) in a script editor or a column formula to create such a table. A quick way to find this information is to select Help > Scripting Index. Change the filter to Functions (top left button) and then select the Probability group. The second list will contain the distribution functions.

Second, you can ask for quantiles after you fit the gamma poisson distribution model. The command is in the red triangle menu for the fit.