Contents

Plots that went wrong

John Lees included in python statistics

2019-09-13 103 words One minute

Contents

There’s loads of these on https://twitter.com/accidental__aRt, here are some of mine from the past couple of years:

/images/intergenic_tchal24.png — The beach?

/images/boxplot_and_violin_idiocy.png — You can never have too many violins

/images/likelihood_surface.png — A nice smooth likelihood surface

/images/nice_contours-1024x745.png — Always good to see the zero contour

/images/terrible_dendrogram.png — ’terrible_dendrogram.png'

/images/isogenic_tau.png — Apply some smoothing then I’m sure it will be fine, right?

/images/regression_separate.png — R2 = 0.11, p < 10E-10

/images/score_contours.png — Let’s cram everything in to 1/10 of the space. 2) The ever so informative labels: ‘1’, ‘2’, ‘3’, …

/images/SPARC_twoComponentBGMM-1024x910.png — Machine learning (or would we call this AI now?). Bonus: the large white bit at the bottom which I couldn’t get rid of

/images/perfect_separation.png — Pong

/images/classification_error_0.png — Cheap Rothko?

/images/classification_error.png — Sheet metal