Add a Pluto notebook to docs

[dmbates]

• Add a Pluto notebook in the docs directory.
• An advantage of Pluto for this is that you can change the settings, hit <shift>-<enter> and have the simulation redone.

[codecov[bot]]

Codecov Report

:exclamation: No coverage uploaded for pull request base (main@23ea639). Click here to learn what that means. The diff coverage is n/a.

:exclamation: Current head 055a76f differs from pull request most recent head e5ee12a. Consider uploading reports for the commit e5ee12a to get more accurate results

@@           Coverage Diff           @@
##             main      #34   +/-   ##
=======================================
  Coverage        ?   93.24%           
=======================================
  Files           ?        4           
  Lines           ?       74           
  Branches        ?        0           
=======================================
  Hits            ?       69           
  Misses          ?        5           
  Partials        ?        0           

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 23ea639...e5ee12a. Read the comment docs.

[dmbates]

@palday Do you think we should keep the section that activates a tempdir? It is used in some of the examples. I assume this is to avoid cluttering up a global environment. In our case it might be best to instruct the user to install JSServer, WGLMakie, etc. before firing up the notebook. I'm not sure how much overhead adding those packages to a fresh environment entails.

[palday]

This is exactly what I've been toying with. I think some of the issues I'm having are related to precompilation issues that I don't seem to have outside of Pluto. I guess it would be good to have a 'normal' Windows and a 'normal' Mac user try out both routes and comment which is easier.

[dmbates]

This is exactly what I've been toying with. I think some of the issues I'm having are related to precompilation issues that I don't seem to have outside of Pluto. I guess it would be good to have a 'normal' Windows and a 'normal' Mac user try out both routes and comment which is easier.

I agree.

I have several questions:

1. I like the idea of using Pluto and the AlgebraOfGraphics/Makie combination. Do you think it will be worth the hassle for the typical user? To me there is a bit of setup fuss but after that I think it is more-or-less self explanatory.
2. What output would the target audience be interested in? For me, having intervals and density plots is more informative than getting p-values and power results but that may be a bit too much tilting at the windmills of students' backgrounds.

[palday]

I like the idea of using Pluto and the AlgebraOfGraphics/Makie combination. Do you think it will be worth the hassle for the typical user? To me there is a bit of setup fuss but after that I think it is more-or-less self explanatory.

I think this is the future of graphics in Julia. It's not yet full ggplot2 level, but basic plots work and more advanced plots can be assembled via Makie primitives. (And doing so is no more effort than the equivalent steps with Gadfly or Plots).

What output would the target audience be interested in? For me, having intervals and density plots is more informative than getting p-values and power results but that may be a bit too much tilting at the windmills of students' backgrounds.

I think power is definitely something to have, but it might be worthwhile to use the density plots showing what that actually means. Might be a good point to review what Type I and II error are. I also really like inference based on intervals + an a priori precision goal (John Kruschke describes his take on this as "the Bayesian New Statistics", but the proposal could be frequentist).

[palday]

@dmbates Still tinkering here or should we I review and merge?

[dmbates]

Still tinkering. I don't expect this branch to be merged - I am just trying out some approaches.