101–104 of 104 posts
Earlier quoted context omitted.
"Missing confidence intervals" is my (IMO) legit pet peeve. People fit frighteningly complicated models with millions-to-billions of parameters. They throw in hairy regularization schemes justified by fancy math. When it comes time to evaluate the output, however: "Our model is better because this one number is bigger than their number [once, on this one test set]."
Confidence intervals aren't even that informative. Like using boxplots when you could inform the viewer so much more with sina plots [1]. Why not show me the whole posterior probability distribution (perhaps helpfully marking the 95% highest density region [2])? Or if you don't have a distribution, show me the 95%, 97.5% and 99.5% intervals. [1] https://clauswilke.com/dataviz/boxplots-violins.html [2] https://www.sci…
It is super weird that an field devoted to doing inference somehow just...doesn't when it comes to evaluating their/our own work.
As an engineer supporting researchers I ran into a peculiar problem. If I suggested an idea they often HAD to ignore it, because they couldn't claim it as their own. Especially the grad students who were trying to get a PhD by coming up with a unique idea of their own, but also the profs because it would be an admission that they "the experts" didn't understand what they were doing. So I had to lead them to come roun…
I really think this misses the mark. Grad students don't treat their prof as infallible - at least good labs. In most labs, the grad students are the ones actually doing the work and research. The prof is just the marketing man. And a good prof in CS will understand they're the clueless marketing man. It's not really feasible for them to keep up with writing code and the whole grant writing (and networking with compa…
As an engineer supporting researchers I ran into a peculiar problem. If I suggested an idea they often HAD to ignore it, because they couldn't claim it as their own. Especially the grad students who were trying to get a PhD by coming up with a unique idea of their own, but also the profs because it would be an admission that they "the experts" didn't understand what they were doing. So I had to lead them to come roun…
> If I suggested an idea they often HAD to ignore it, because they couldn't claim it as their own. Especially the grad students who were trying to get a PhD by coming up with a unique idea of their own, but also the profs because it would be an admission that they "the experts" didn't understand what they were doing. That's what authorship is for. What's the point in supporting academia if you can't get authorship?