Layout
In plotscaper: Explore Your Data with Interactive Figures

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Adding and removing plots

As you have probably seen in other vignettes, you can add plots using the various add_plot functions:

library(plotscaper)

names(airquality) <- c("ozone", "solar radiation", "wind", 
                       "temperature", "month", "day")

schema <- create_schema(airquality) |>
  add_scatterplot(c("solar radiation", "ozone")) |>
  add_barplot(c("day")) |>
  add_histogram(c("wind")) |>
  add_pcoords(names(airquality))

schema |> render()

You can also remove plots:

schema |> 
  remove_plot("bar1") |> # Remove the first barplot added to the figure
  pop_plot() |> # Remove the last plot added to the figure - typically, the bottom-right one
  render()

This does not make much sense server-side, since, if you don't want a specific plot to be in the figure, you can just not add the figure in the first place. However, on the client-side, this can be very useful, since you can modify the figure on the fly:

#| eval: false
# NOT RUN - this only makes sense inside a running R session
scene <- create_schema(airquality) |> render()

scene |> add_scatterplot(c("solar radiation", "ozone"))
scene |> add_barplot(c("day")) # Oops, maybe I want a histogram instead
scene |> pop_plot()
scene |> add_histogram(c("wind"))

Some of this was already discussed in the Get started section.

Layout

We can also modify the layout of a plotscaper figure, using an interface that's very similar to graphics::layout. If you're not familiar with this function, we can provide it a numeric matrix, with values identifying plotting regions as contiguous rectangles, and the plots will be appropriately resized:

# 1. Big scatterplot on the top left
# 2. Small barplot on the top right
# 3. Small histogram on the right
# 4. Wide parallel coordinates plot on the bottom
# 0. (empty space on the bottom right)

layout <- matrix(c(
  1, 1, 2,
  1, 1, 3,
  4, 4, 0
), ncol = 3, byrow = TRUE)

schema |> set_layout(layout) |> render()

The matrix can be of any size, the only requirement is that the plot areas must be contiguous, meaning that e.g. this would not be a valid layout:

#| eval: false

layout <- matrix(c(
  2, 1, # Cannot split plotting regions like this
  1, 2
), ncol = 2)

Also, all of the plots should be included in the layout. If they are not, a pop-up warning will show up. You can decide whether you want to proceed with the figure, however, be warned that the layout may not work well.

Finally, you can call the set_layout function from inside a running R session with scene |> set_layout(...) (and you will get a client-side update to the current figure).