Choose Language Hide Translation Bar
Highlighted
NoScoped
Level I

Quantile regression with overdispersed count data?

What's the best way to perform a quantile regression with overdispersed count data?

4 REPLIES 4
Jeff_Perkinson
Community Manager Community Manager

Re: Quantile regression with overdispersed count data?

We'll probably need some more information about what you're facing but the Generalized Regression personality in Fit Model in JMP Pro will do Quantile Regression. 

 

Can you tell us what you've tried and what your data is?

-Jeff
Highlighted
NoScoped
Level I

Re: Quantile regression with overdispersed count data?

My data are egg counts (i.e., anywhere from 0 to ~35,000). I'm currently using a Poisson GLM with an overdispersion parameter. I'm simplifying here (and I'm sure I'm glazing over some important details) but since my response variable can never be less than zero, one of my committee members has recommended that I try running a quantile regression. 

 

I'd be happy to go into more detail about my experimental design if they would help provide some context for my data.

Highlighted
Jeff_Perkinson
Community Manager Community Manager

Re: Quantile regression with overdispersed count data?

Quantile Regression from the Generalized Regression platform in JMP Pro is totally reasonable in your case.

 

You could also use negative binomial regression, but either way is probably fine.

-Jeff
Highlighted
P_Bartell
Level VI

Re: Quantile regression with overdispersed count data?

And to add a bit to @Jeff_Perkinson 's advice, since you are going to be using JMP Pro, I encourage you to leverage JMP Pro's Model Comparison platform to help you visualize and decide which modeling approach works best for the practical problem at hand. Here's a link to the JMP Pro online documentation for the Model Comparison platform:

 

https://www.jmp.com/support/help/14-2/model-comparison.shtml#

 

Article Labels

    There are no labels assigned to this post.