This is madness. 60% accuracy at best on your training dataset (2019 data)! So it will be in all likelihood worse on the 2020 data. How did anyone approve this to be used. And it's not like it's important to a whole heap of people's future. Awful.

from Twitter


Popular posts from this blog

A census of amplified and overexpressed human cancer genes : Nature Reviews Cancer

RT @jburnmurdoch: Most useful thread I’ve read on the dynamics of Covid transmission. By an expert, and packed full of links to evidence. Indoor spaces and or prolonged face-to-face contact account for large majority of Covid transmission. Walking past someone in street/park much lower risk.

GMail now with IMAP