I realise the benchmarks are still drafts, but I had fun implementing them in R's Tidyverse anyway, which is the most expressive data analysis/wrangling DSL I know. Pic and source attached, should somebody find it useful.

@Srini K more broadly, I’m searching for a taxonomy of tabular analytics tasks. I didn’t have a particular method for selecting tasks. I just picked what seemed common to my personal experience. But I’d like a more systematic way of picking tasks that cover the space. For example, this paper is a good start: https://ieeexplore.ieee.org/document/1532136