Open DivadNojnarg opened 3 months ago
Just for fun
I like the idea, this is deliberately engaged and some part of truth in it. Nice opening hook.
From Karma: emphasise flexibility of blockr for EDA From Karma, Nicolas: display how blockr leverages the power of R (stats, modeling, etc.)
@JohnCoene A summary of the recent progress.
I did a draft repetition of the presentation yesterday including the "Easy Not so easy web apps for data Science without the compromise" slide as you suggested. People liked the comparison between how much code it takes to create an app with blockr vs Shiny.
Regarding the slides content, I drafted the intro, first part (general concepts), extend (create block + registry). For the examples, I did something with the palmer penguins as I had to have something working for the first presentation draft (I had issues with your old faithful example as I get an invisible plot. I you recall, we had this issue with the scatterplot block). I realise we have a lot to say and we might end up with issues if we go for more complex cases. TODO: discuss about what we show in the demo: do we go for simple but not so impressive example or more complex but maybe overwelhming example?
For instance, I'd like to show the dashboard you developed in blockr.app
to see how far we can go in term of customisation/design. Maybe something which can do at the end of the "extend" part when we talk about the customisation... Another challenge is the time it might take (I recall Karma spending 5-10min for the demo which is too long for our talk).
In the slides, I still don't mention the serialisation/restore feature. If you want/can take care of this at the end of the first part, after the workspace slide. That's important for the reproducibility.
TO DO: update blockr.ggplot2
and others with latest blockr changes for the demo apps.
TO DO: who does what. How do we transition between part.
@JohnCoene : preliminary split
Slide 1 -> David: bad cooking experience Slide 2/3/4 -> John: Shiny's is easy until its not ... + Talk about BMS contacting us to create analysis block by block. Slide 5 -> David: blockr reveal
Slide 6/7/8 -> David: ggplot example, recipe metaphor, stack definition. Slide 9: blockr first demo -> John TO DO: add filter block (maybe remove the submit button option so we don't have to click on it ...) Slide 10 -> John: how much time it would take with Shiny? Slide 11 -> John: How much code with blockr ? Slide 12/13/14/15 -> David: Workspace intro + demo + transition to part Extend
John: intro + code slides David: registry + customize if time.
Note: not sure we have time to talk about customisation. Maybe we need to open on this in the conclusion slide dedicated to advanced developers?
@JohnCoene I replace the first demo (shinylive) by a video recording. For that, I worked on the add-block branch (plus merging main inside to get the latest changes). I'll do the same for the workspace.
I'll move the shinylive demo to the appendix as we might need them during the questions.
cc @JohnCoene
Intro/Opening:
blockr 101: basic definition and usage
Extend the supermarket: Blocks are gathered in a registry, which can be extended.
Towards a COOP system: connect markets together (we're not there yet but that's our long term goal)
Conclusion/Call to action: