TidyTuesday with Fable

A series of exploratory data analysis sessions on TidyTuesday datasets, produced by Jon Minton and Claude (Fable 5) — in two distinct working modes.

The sessions come in two kinds, and the site keeps them clearly separated because the way a page was made is part of what it is:

There’s a short reflection on the two modes comparing how they felt and what came out differently.

Session 2 · autonomously developed

Fable’s own picks, analyses and designs — one page in each of four registers.

  • How likely is ‘likely’? (statistical inference) (TidyTuesday 2026-03-10) — 5,174 people put numbers on 19 probability phrases. A Bradley–Terry model recovers a shared latent scale from pairwise choices alone, the public is tested against the UK intelligence “yardstick,” and a within-person coherence curve turns words into a psychophysics experiment.
  • Thirteen males per female: an island running out of mothers (substantive demography) (TidyTuesday 2026-03-03) — sixteen years of capture data on the Hermann’s tortoises of Golem Grad, where sexual harassment, emaciation and tiny clutches compound into a projected extinction. The mechanism of a “demographic suicide,” measurement by measurement.
  • A million digits of π, and not a pattern in sight (novel visuals + randomness theory) (TidyTuesday 2026-03-24) — a turtle walk through 100,000 digits, a convergence funnel, a digit-pair heatmap and the Feynman point: every test for pattern comes back empty, on a number that is the opposite of random.
  • How a cold sea breathes (physical time series) (TidyTuesday 2026-03-31) — eight years of a Nova Scotia water column as a depth–time heatmap: the annual heat pulse damping and lagging as it sinks, and the winter weeks when the sea turns upside-down.

Session 1 · co-developed

Built live, in conversation — the original sessions.