Great stuff. I've found these threads a fun and fascinating ML, stats and geophys. tutorial.
I don't know which ML and regression graphs you're going to present but, apart from the 1980-1996 biennial flip, I think I'd find a bullet point list of the different time periods for different judgements would obviate any cherry picking doubts. Was it the Astrudillo et al. paper which first posulated the flip? Causes of the flip seems like a great sign off challenge for an audience.