test

r11

zhufei

Friday, May 16, 2014

First we load ggvis and shiny:

options(width = 200)
library(ggvis)
library(shiny)

Then we create a linked brush object. This needs to know the possible linking keys. Here we’re going to link together some plots of the mtcars dataset, so we’ll add an id column, and use it as the keys:

# Add ID column to mtcars so that we can select from it
mtcars2 <- cbind(mtcars, id = seq_len(nrow(mtcars)))

lb <- linked_brush(keys = mtcars2$id, "red")

Next we create two plots. Points that are brushed on the first plot will be displayed on the second plot, and the scales will automatically adjust so that those points fill the plot.

mtcars2 %>%
  ggvis(~disp, ~mpg, key := ~id) %>%
  layer_points(fill := lb$fill, fill.brush := "red") %>%
  function(vis) lb$input(vis)

## Warning: Can't output dynamic/interactive ggvis plots in a knitr document.
## Generating a static (non-dynamic, non-interactive) version of the plot.

# A subset of mtcars2, of only the selected points, or the whole data if nothing
# is selected.
selected <- lb$selected
mtcars2_selected <- reactive({
  if (!any(selected())) return(mtcars2)
  mtcars2[selected(), ]
})

mtcars2_selected %>%
  ggvis(~disp, ~mpg) %>%
  layer_points(key := ~id) %>%
  set_options(duration = 100)

## Warning: Can't output dynamic/interactive ggvis plots in a knitr document.
## Generating a static (non-dynamic, non-interactive) version of the plot.

<U+269

• Renderer: SVG | Canvas
• Download

<U+269

• Renderer: SVG | Canvas
• Download

Also, we can show a table of the data, including a column indicating which ones are selected:

renderTable(
  cbind(mtcars2, selected = selected())
)