feedback

[DominiqueMakowski]

@strengejacke @pdwaggoner @mattansb @humanfactors @IndrajeetPatil

I know it's summer and all, but if you wanna really relax at the beach, enjoying a nice Pina colada and a good read, I can only suggest the following vignettes, fresh out of the oven, showcasing some of the features of estimate 😁:

• Data grids
• Marginal means
• Contrast analysis
• Use a model to make predictions
• Describe non-linear curves

In fact, for me this package is pretty much ready for a first release (right after parameters is accepted). Although supporting mainly rstanarm models, for now, once we have the API right it will be easy to extend (in fact, estimate_means and estimate_contrasts already support base/lme4 models 🤐).

Anyway, please do not hesitate to give your feedback on any aspects of the package!

[mattansb]

Looks really great! Seeing as how these vignettes are very visual, was wondering... how could @strengejacke ggeffects fit into all of this? 🤔

[strengejacke]

After a quick look, I would say that the functions of ggeffects do what estimate_response() does with data = "grid". estimate_means() is comparable to the margins-package, and estimate_contrasts() looks like what emmeans::pairwise() can do.

We've already thought of moving ggeffects over here to easystats (https://github.com/easystats/easystats/issues/15), but hesitated to do so because it did not fit perfectly into the scope back then, and it does not strictly follow the naming-conventions we have. But I do think that there'll be some overlap between estimate and ggeffects, and probably estimate can benefit from the experiences I have made with ggeffects ;-)

[DominiqueMakowski]

agreed and moreover, for now ggeffects supports 10000 times more packages... for now 😈