based_on not working in @linq macro about dataframesmeta.jl HOT 5 CLOSED

Looks like only the @based_on form (with the @) works:

iris=dataset("datasets", "iris")
@linq iris |> groupby(:Species) |> @based_on(x = mean(:SepalWidth))

But this syntax is probably better, right?

@linq iris |> by(:Species, x=mean(:SepalWidth))

Or maybe I'm missing the difference between by and based_on (@tshort again, do we need both?).

from dataframesmeta.jl.

If you want to do transformations or calculations in between grouping and summarizing, a sand-alone based_on is nice. For an example, say you've got a data frame with the positions (x) of a bunch of different objects at different times (t), and you want to calculate each object's average speed:

speeds = @linq trajectory |>
    group_by(:object) |>
    transform(timespan = max(:t) - min(:t), displacement = last(:x) - first(:x)) |>
    based_on(avg_speed = :displacement ./ :timespan)

I this could of course all be done on one line, or by defining a helper function, but if you're working in the data-frame-piping style I think a separate based_on or summarise makes things a little clearer...

from dataframesmeta.jl.

👍

from dataframesmeta.jl.

OK, I see. Though I don't understand the name: what is "based on" what in this operation? Isn't summarise a better term?

from dataframesmeta.jl.

I believe the ideas is "based on this data frame, calculate these things." I'd tend to agree that summarize is a better term. In dplyr, all the data frame operations are verbs, which is nice for readability.

(Prepare for the coming flame war over whether it should be summarise or summarize ;) )

from dataframesmeta.jl.