fitted functionality

[zacksteel]

Hi Jeff,

Thanks again for a great package. I have a suggestion/request for additional functionality. The various model types can be passed through the predict function to get posterior predictions, which is very handy. However, it would also be nice to be able to easily get fitted/expected values in a similar fashion (e.g., to plot marginal effects like this). I see that you use the generic fitted function as part of the goodness of fit assessments, but this produces y and p matrices but not z or psi. Would it be doable to add a convenience function that also produces expected values of the latent state?

Cheers,
Zack

[doserjef]

Hi Zack,

Thanks for the note and sorry for the delay in response. The posterior distributions for the latent states z and psi (for the sites used to fit the model) are provided directly in the model fit object. So if you saved the model results in an object called out, you can access those values directly via out$psi.samples and out$z.samples. It is on my todo list to make easier functions for generating marginal effects plots to make those a bit easier to generate (it's a bit clunky now, but here is a repository that contains some code with examples of doing that). Thanks again for the suggestions, I'll keep this issue open for now until I can get some nicer functions together for making marginal effects plots.

Thanks!

Jeff

[zacksteel]

Thanks Jeff. I'm sure you've got plenty on your to do list for this package! What you've done already is great and I'll look forward to seeing its continued development. Zack

…

On Fri, Apr 12, 2024 at 8:00 AM Jeff Doser ***@***.***> wrote: Hi Zack, Thanks for the note and sorry for the delay in response. The posterior distributions for the latent states z and psi (for the sites used to fit the model) are provided directly in the model fit object. So if you saved the model results in an object called out, you can access those values directly via out$psi.samples and out$z.samples. It is on my todo list to make easier functions for generating marginal effects plots to make those a bit easier to generate (it's a bit clunky now, but here is a repository <https://github.com/eco4cast/Statistical-Methods-Seminar-Series/tree/main/doser-finley-spocc> that contains some code with examples of doing that). Thanks again for the suggestions, I'll keep this issue open for now until I can get some nicer functions together for making marginal effects plots. Thanks! Jeff — Reply to this email directly, view it on GitHub <#42 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEYO5FGG46XIR4OX55M6FT3Y47SH3AVCNFSM6AAAAABF3OFSX6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANJRHAYTSNRRGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>