Hello all, I am working on a project that is modeling life expectancies for several countries and years. We are trying to assess model uncertainty of an estimate using a Bayesian approach (e.g., the effect of GDP on life expectancy) not only regarding the specification of a particular class of mod…

By “OLS” do you mean linear regression? Usually optimization with least squares doesn’t give you uncertainty. [image] sdaza: I could combine those distributions weighting them by a performance measure. I’m not sure what you mean here. You can, in general, define derived quantities in terms…

Thanks for your answer, Bob. OLS = linear regression. Sorry, my question was no clear. What we are trying to do is some kind of model averaging using different classes of models (linear, non-linear, boxcox). I haven’t seen applications of that kind of model averaging or stacking using a Bayesian a…

[image] sdaza: Sorry, my question was no clear. What we are trying to do is some kind of model averaging using different classes of models (linear, non-linear, boxcox). I haven’t seen applications of that kind of model averaging or stacking using a Bayesian approach. How about Bayesian stacki…

Thanks for your reply! That means that if do something like: m1: y ~ x1+ x2 m2: log(y) ~ x1 + x2 # dependent variable transformation Can I get model weights using? log_lik_list = list() log_lik_list[[1]] = extract(m1)[[“log_lik”]] } log_lik_list[[2]] = extract(m2)[[“log_lik”]] } loo::model_…

[image] sdaza: Any ideas or suggestions? You can just make a mixture model.

[image] Bob_Carpenter: You can just make a mixture model. Stacking paper section 4.3 demonstrates that mixture model with small data can perform badly (at least in M-open case).

[image] avehtari: Stacking paper section 4.3 demonstrates that mixture model with small data can perform badly Thanks for the pointer. I’ll take a look.

Model uncertainty using Stan

General

avehtari February 7, 2018, 6:23pm 6

No. You need to take into account the Jacobian of that transformation log(y) when computing the log predictive density. See my comment in thread Model comparison via PSIS-LOO for 3-level hierarchical growth model - #4 by avehtari

Topic		Replies	Views
Model averaged parameter estimates Modeling loo	3	1112	February 9, 2018
Calculating model averaged estimates using loo Modeling loo	8	1375	January 31, 2024
Loo_compare vs averaging/weighting via stacking or pseudo-BMA weighting rstanarm fitting-issues , loo	2	973	February 5, 2021
Question on model stacking -- averaging posteriors of parameters, not predictions Modeling loo , model-comparison	19	1939	July 24, 2025
Implementing Model Averaging RStan loo	9	3877	August 17, 2022

Model uncertainty using Stan

Related topics