From 98ec2e3882b4a4e9e4c6a73ed78e40970041e901 Mon Sep 17 00:00:00 2001 From: Jose Storopoli Date: Fri, 29 Dec 2023 16:09:07 -0300 Subject: [PATCH] fix(04-workflow): typst format --- slides/04-bayesian_workflow.typ | 6 ------ 1 file changed, 6 deletions(-) diff --git a/slides/04-bayesian_workflow.typ b/slides/04-bayesian_workflow.typ index 63c8fb3..b87a326 100644 --- a/slides/04-bayesian_workflow.typ +++ b/slides/04-bayesian_workflow.typ @@ -125,8 +125,6 @@ We need to make sure that the posterior distribution of $bold(y)$, namely $bold(tilde(y))$, can capture all the nuances of the real distribution density/mass of $bold(y)$. - #v(1em) - This procedure is called *posterior predictive check*, and it is generally carried on by a visual inspection #footnote[ we also perform mathematical/exact inspections, see the section on _Model Comparison_. @@ -134,14 +132,10 @@ of the real density/mass of $bold(y)$ against generated samples of $bold(y)$ by the Bayesian model. - #v(1em) - The purpose is to compare the histogram of the dependent variable $bold(y)$ against the histograms of simulated dependent variables $bold(y)_"rep"$ by the model after parameter inference. - #v(1em) - The idea is that the real and simulated histograms blend together and we do not observer any divergences. ]