Learn how to make your own Observable notebooks with these guided walkthroughs.


Paul Buffa

Allison Horst

Robert Kosara

Josh Earlenbaugh

Observable

Hands-on tutorials

Learn Just Enough JavaScript: Introduction

Observable Quick Guides

Build your first scatterplot with Observable Plot

Get started with the Chart cell

Build your first map with Observable Plot

Build your first choropleth map with Observable Plot

Plot from ggplot2

Ian Johnson

Analyzing revenue by category for an online store

Which items are purchased together?

Wrangling data in JavaScript with Arquero: a primer for R users

New to Observable but have a background in Excel? This series is for you. 

Observable for Excel Users

Getting started with Observable? You’ve come to the right place.

Tutorials

Hands-on Tutorials

Build and advanced report with simple components. Starting with a SQL query, add interactive charts and dashboard controls.

Which items are purchased together

I think it might help to talk here more about co-occurance, why it's important, maybe some examples of it used in the wild?

do you have ideas where to find examples? i was kind of hoping this would be an example. i agree on giving a short overview

Maybe @allisonhorst you have some suggestions? I know its used a lot in the NLP space, but I don't have a great go-to example on hand...

Here's a recent example on co-occurrence of Covid-19 symptoms: https://kieranhealy.org/blog/archives/2020/04/16/upset-plots/. I think this nicely highlights failures of Venn Diagrams, so could be useful in setting up clearer ways to visualize co-occurrence w/ more complex datasets. + I love a good UpSet plot.

It might help with accessibility to add that Plot colorContrast helper - not sure though if you want to add that level of complexity though. 

https://observablehq.com/@observablehq/plot-colorcontrast-custom-transform?collection=@observablehq/plot-examples

actually that's a really nice addition to the matrix visualization tutorial (importing an improvement and applying it)

I think it might be good to start the graph with a default filter, one that doesn't have so much overflow. (I also think there's ways we could tweak the simulation parameters to make this happen less.)