Hello all, I am implementing the Besag York Mollié (BYM) spatial model as specified by Mitzi Morris in the iCAR case-study. https://mc-stan.org/users/documentation/case-studies/icar_stan.html There is plenty of background at that link, but in general it is a Poisson GLM with both an ordinary rando…

OK, on second thought - your case is likely different, as you transfer to completely new territory, not just new districts in the same territory. In your case, I would first stop to think what question are you answering. Since you cannot ground your predictions for Boston in any actual data from Bo…

[image] martinmodrak: I would first stop to think what question are you answering. That is the correct answer to most of my questions! Thank you for your time and thoughts. My interest can be summed up to, “If we can model the count of incidence (y) conditioned on the environmental factors (b…

[image] mrecos: However, rho is expressed via the convolved_re vector, and that depends on phi which needs known incidence counts y to make the node1 and node2 Maybe I am missing something (I have very superficial knowledge of iCAR), but does it really? Imagine this process: Take a sample o…

BYM2 spatial model - Making predictions to new area/geography from fit model

Modeling

martinmodrak March 18, 2019, 12:36pm 2

I think I semi-answered a similar question recently: Sparse CAR predictions for hold-out data Does that apply to your situation?

Topic		Replies	Views
BYM2 with unstructured country random effect Modeling	39	2776	February 11, 2022
Bym and besag models in brms brms	34	3421	September 17, 2019
Bym2 General rstan , fitting-issues , cmdstanr , hierarchical-model , brms	26	1622	February 14, 2023
Bym model General stan	11	1386	August 26, 2022
Aggregate predictions of brms::posterior_predict brms	1	950	August 13, 2018

BYM2 spatial model - Making predictions to new area/geography from fit model

Related topics