Yuling Yao

> [@axch] input files: Rather than hardcoding synthetic inputs, I think it would be better to deliver a reproducible generator program (i.e., one that gives the same results when running...

>I don't understand the difference. A prediction is coded as a parameter in Stan (or a generated quantity if it's simple enough and can be factored). Either way it's a...

Would Pr(k>0.7) be a more reliable diagnosis with small MC sample size? On Mon, Sep 13, 2021 at 3:28 PM Aki Vehtari ***@***.***> wrote: > A few times users have...

Perhaps in this case the loo package can serves as a stan-model-generator: compute pointwise loo, suggest a default stacking model, and generate a stan file. Then user can run/modify this...

Hi Staffan, If you have a weighted likelihood e.g. target + = frequency_weight[i]* likelihood [i], then I think you can still treat it as a vanilia exchangebable model--the only difference...

Some math: ### CRPS score The CRPS for forecast CDF $F$ is $$\mathrm{crps}(F,y)=E_X|X-y|- \frac{1}{2}E_{X,X^\prime}|X-X^\prime|.$$ where $X$ and $X'$ are two independent copies from $F$. Using these predictive draws $z_{m1}, \dots,...

@LeeviLindgren I might have misunderstood the shape in your original code: `crps.default

@LeeviLindgren the separation for the matrix and vector looks good. One more question: when computing Exx, in theory we can average over all permutations to reduce variance (In the current...

I thought delta_ELBO_mean and delta_ELBO_med have already been printed?

I do not know why I cannot get assigned or self-assigned. But yes, I am working on this. I have implemented this in cmdstan and will try to add it...