jasp-stats / jasplearnbayes Goto Github PK

e.g., sequential updating

Found this typo while translating

Currently, the LSgameofskill.R allows the user to set the PSP of all players to 0, which makes no sense under a beta-binomial distribution.

I just took a more thorough look at the Learn Bayes module, for my upcoming teaching tasks (intro Bayes first year bachelors students). I noticed two possible improvements:

• for Binomial estimation and testing, the first menu is "Data", and after that comes "Model" or "Hypothesis". I would suggest turning these around, since ideally first we think about which models or hypotheses we want to test, and how we want to specify the prior distribution. This also makes sense for the output panel, since we need to specify a model or hypothesis for anything to show up there (whereas this is not the case for data - the output is there even if the data are left at 0 and 0).
• I am missing the classic JASP prior/posterior plot, with both distributions, grey dots indicating the Savage-Dickey ratio, and the Bayes factor, all in one figure. I think having this figure is a lot more concise than the current way of providing many different plots (hampers the overview a lot, imo). This might be a challenge, since the idea in the module is to just have models, rather than an explicit H0 and H1 (although we could use the same approach as used under "Bayes factor" under "Predictive Performance", where you select the vs model in the dropdown menu).

What do you think @EJWagenmakers and @FBartos?

The binomial estimation learnbayes module does not update when changing the observed count data.
See screenshot below - where data are added, but posterior is not updated:

commit id: jasp-stats/jasp-desktop@8f9be0d