Interesting, can you recommend a resource on this?

While stata/r/python will get slower in these things the bigger the data, power BI scales very well.
Try loading a 3 GB csv with 50 variables, some string, into stata/python/r. dta file will be 500 MB, power BI file will be 50 MB. When I had 100GB once, power BI file was 100 MB.

What takes indeed time is changing the data import settings (because they need to rerun their compression algorithm?).

But once the data set(s) are compressed it is comparatively fast to do any analysis and browse data.

I found it useful for exploring these bigger data sets and presenting intermediate results, but maintained the power BI files separately from the main pipeline that produced the final results. That worked.

Indeed ugly and not built for research (so many useless buttons🫨).
It‘s free, though, and compression is insane. Easily handles data sets >>10GB, e.g., transaction level data – even on a standard local machine.

(*fiddles)

It can, although it turned out it does not like`titlesec` (github.com/quarto-dev/q...)

This was the resulting pdf:
github.com/moritz-schwa...
I only spot some differences in the kerning.

Congrats to the 4 additional cites since the last compilation :)

What I was looking for! Probably runs in quarto too (with minimal adjustments)?