Enhancement request: GUI for a model editor for Bayes data analysis

jasp-stats / jasp-issues

This repository is solely meant for reporting of bugs, feature requests and other issues in JASP.

57 stars 29 forks source link

Enhancement request: GUI for a model editor for Bayes data analysis #118

Closed fplatz closed 5 months ago

fplatz commented 6 years ago

Enhancement: Statistical Modelling
Purpose: model editor for Bayer analysis This is my last post for today (I promise you that!): Is it possible to develop a GUI for building statistical routines and models as proposed by Lee and Wagenmakers (2014) or Kruschke (2014) but not by typing endlessly lines of (jag of wing) code but instead by using icons (for distributions) arrows etc. as the authors used to visualize the bayes alternatives to the NHST approaches. The only thing a scientist have to do is to give the distribution parameters or mathematical formulas (at a minimum). Perhaps this might be a step towards a new way of click and Drag & Drop - Bayes data analysis/modelling?

Cheers, Friedrich

EJWagenmakers commented 6 years ago

We are working on more generic methods, but a GUI for building models is probably not forthcoming. The reason is the disaster called "DoodleBUGS", which is the Clippy of probabilistic programming languages. Of course it is a good idea in principle (which is why DoodleBUGS was developed), but it just didn't catch on at all. What would be interesting is to find a way to automatically produce the Doodle from the model syntax, but I think that might be tricky.

tomtomme commented 5 months ago

@fplatz @EJWagenmakers Any update on the horizon for bayesian model editors after 7 years? Or should we close this one for now.

fplatz commented 5 months ago

I think, that this idea is of high value not only for research but also for educational purposes. But I understand, that this might not be of main interest.Am 20.02.2024 um 11:40 schrieb Thomas Langkamp @.***>: @fplatz @EJWagenmakers Any update on the horizon for bayesian model editors after 7 years? Or should we close this one for now.

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: @.***>

EJWagenmakers commented 5 months ago

Well I will say that there are popular GUIs for Bayesian expert networks. All of this can be done in JAGS/Stan, but the GUIs remain very popular. So I can see the added value there, but it would be a tremendous effort. So I don't think it is a high priority and since we haven't done anything about it in 7 years I will close this isse.

fplatz commented 5 months ago

I can understand this point of view. When I suggested this idea, I thought if an „all-in-one-box“-solution as this seemed to me as overall mission for JASP due to the fact, that several solutions did also exist before JASP. But I see, that this is to resource demanding.Anyway, please keep on moving JASP as you did! It’s a fantastic solution! Am 20.02.2024 um 12:07 schrieb EJ @.***>: Well I will say that there are popular GUIs for Bayesian expert networks. All of this can be done in JAGS/Stan, but the GUIs remain very popular. So I can see the added value there, but it would be a tremendous effort. So I don't think it is a high priority and since we haven't done anything about it in 7 years I will close this isse.

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: @.***>