Post

group_by() + summarize() too old to switch to the other approach

2. The way it's written within a piped sequence is opposite to the order it was: reading top to bottom left to right we compare group=>summarize to summarize=>group. I'd rather write out summarize(.by = group, a = mean(x)) But that's not the order in the function, and also...

I was going to say group_by() but it looks like a bunch of other people beat me to it so to summarize() this discussion I would say group_by()😂

group_by().

Both! I break my summaries in those groups. There’s usually a group_by chunk (ie, summaries of the same group) and a .by chunk (where I want a single pipe with multiple grouped outputs).

Just here to throw tapply() into the mix 😉 (I've also occasionally used group_by)

option a.

For me: `.by`, all the way and in all functions, even in`slice_sample()` where it is annoyingly called `by`. But `.by` deserves its own row.

If I want to use .by, I would need to totally commit to making that my workflow. No group_by(), or at least none without immediately ungroup()ing. And that's a relatively heavy cognitive lift.