References for the above, since these papers are not well known, but should be:

King et al (2017): doi.org/10.1111/ajps...
Huling & Mak (2024): doi.org/10.1515/jci-...
Cho et al (2013): doi.org/10.1111/stan...

It's a legitimate way to measure metabolism. It has higher error than actual direct or indirect calorimetry, but it's "fine" at scale.

Haven't read the rest of paper yet...

Still debating if this is worth a full blog post with reproducible code. Let me know if you'd read it and find it useful...

I have a 6 and 8 year old. English irritates them to no end because of things like this.

I think you mean compromize.

I had someone ask recently if I did a sensitivity analysis recently and literally had to ask "sensitive to what?" If you want a sensitivity analysis to every analytical decision I make, you're in for a long presentation!

Please post when there is a preprint or something citable! I have a paper currently under review where it may be helpful.

I love me a tensor spline interaction.

This was discussed elsewhere, I think it's a philosophical difference.

I'd argue the other way around, how would you know what outcome is meaningful if you don't understand the system? You don't design a drug without a planned cellular target.

Haha, I've heard the purpose of data science is to confirm choices the C-suite have already made...

It's just hard to toe the line between using words normal people understand and being totally wrong in describing what's going on.

😂 It's the C-suite, the last step is always implied.

But yeah, I think my boss just wants something for a PPT saying "look! The magic has a process you understand!" because the actual analysis plan requires a PhD in statistics and lots post doc learning to get.

They requested a figure. I think part of the idea is just to show that we have a process, it takes many steps to do well.

It's accompanied by a 10 page statistical analysis plan they won't likely read or understand if they did.