3 ms·
That is basically R with tidyverse. flights |> filter( carrier == "UA", dest %in% c("IAH", "HOU"), sched_dep_time > 0900, sched_a
by minkles 2y ago
That is basically R with tidyverse.
flights |>
filter(
carrier == "UA",
dest %in% c("IAH", "HOU"),
sched_dep_time > 0900,
sched_arr_time < 2000
) |>
group_by(flight) |>
summarize(
delay = mean(arr_delay, na.rm = TRUE),
cancelled = sum(is.na(arr_delay)),
n = n()
) |>
filter(n > 10)
If you haven't used R, it has some serious data manipulation legs built into it.
- dan-robertson 2y agoAn interesting thing to me about all these dplyr-style syntaxes is that Wickham thinks the group_by operator was a design mistake. In modern dplyr you can often specify a .by on an operation instead. I found switching to this style a pretty easy adjustment, and I think it’s a bit better. Example: d |> filter(id==max(id),.by=orderId) I think PRQL were thinking a bit about ways to avoid a group_by operation and I think what they have is a kind of ‘scoped’ or ‘higher order’ group_by operation which takes your grouping keys and a pipeline and outputs a pipeline step that applies the inner pipeline to each group.
- _Wintermute 2y agoGiven 10 more years dplyr syntax might resemble data.table's
- countrymile 2y agoMy thoughts exactly, it even uses the same pipe syntax, though I do prefer `%>%`. I've been avoiding SQL for a while now as it feels so clunky next to the tidyverse