animint2: An R repository from lazycipher

animint2 provides an animated, interactive grammar of graphics

Introduction

Animint2 is a fork of ggplot2 that makes it possible to design multi-layer, multi-plot, interactive, and possibly animated data visualizations using just a few lines of R code. Animint2 adds the clickSelects and showSelected interactive keywords to the grammar of graphics, and renders to the web using D3. For example, this multi-layer facetted interactive animation of WorldBank data was defined using only about 60 lines of R code.

Installation

Now available on CRAN:

install.packages("animint2")

Usage

The best reference for learning is the Animint2 Designer Manual, which provides several examples and exercises to get you started using animint2.

For some examples of what kinds of interactive data visualizations can be created, see the animint gallery.

For additional usage examples, see our extensive set of test cases.

Differences with other packages

Our JCGS paper provides a detailed comparison (an executive summary is below).

R packages for animated graphics

R packages gganim, animation and gganimate can generate animated graphics, in which the only interaction is going forward or backward in time. In contrast animint2 provides animation and also interaction with other variables (not only time).

shiny+plotly web applications

Typical shiny web applications provide interactive graphics with indirect manipulation (via menus/buttons/etc) but the emphasis in animint2 is direct manipulation (via mouse clicks on lines/points/etc). Direct manipulation in shiny apps can also be achieved using the excellent plotly package, in which interactivity can be specified using events; in contrast animint2 uses the clickSelects/showSelected keywords to specify interactivity.

loon exploratory graphics

loon is an excellent R package for interactive exploratory graphics; the focus of animint2 is interactive presentation graphics.

Old animint

animint2 is a redesign of animint with:

Cleaner syntax. In the old animint we had showSelected/clickSelects as aesthetics, and in animint2 they are now geom parameters.
Easier installation. The old animint depended on ggplot2 but animint2 does not (it has copied the necessary functions from ggplot2).

For a concrete example of how the syntax changed, consider the following example dataviz, adapted from tests/testthat/test-renderer1-variable-value.R. The data set used to draw the blue line segments in the bottom plot looks like this:

> with(peak.problems, data.frame(selector.name=paste0(problem.name, "peaks"), problem.name, peaks, bases.per.problem))
             selector.name       problem.name peaks bases.per.problem
1  size.100.problem.1peaks size.100.problem.1     1               100
2  size.100.problem.2peaks size.100.problem.2     1               100
3   size.50.problem.1peaks  size.50.problem.1     1                50
4   size.50.problem.2peaks  size.50.problem.2     1                50
5   size.50.problem.3peaks  size.50.problem.3     1                50
6   size.50.problem.4peaks  size.50.problem.4     1                50
7  size.100.problem.1peaks size.100.problem.1     2               100
8  size.100.problem.2peaks size.100.problem.2     2               100
9   size.50.problem.1peaks  size.50.problem.1     2                50
10  size.50.problem.2peaks  size.50.problem.2     2                50
11  size.50.problem.3peaks  size.50.problem.3     2                50
12  size.50.problem.4peaks  size.50.problem.4     2                50
>

and the old animint code looks like this:

geom_segment(aes(
  showSelected.variable=selector.name,
  showSelected.value=peaks,
  clickSelects=problem.name,
  showSelected2=bases.per.problem),
  data=peaks.dt)

In both animint and animint2, there are “selectors” which are variables in the interactive graphic that can change based on what you click on. In the old animint, selectors were specified using aesthetics:

The aes(clickSelects) means that whenever you click on one of these segments, the problem.name selector will change. For example clicking the segment that is plotted for the first row of data will change problem.name to size.100.problem.1.
The aes(showSelected2) means that the only segments that will be shown are the ones which correspond to the current value of the bases.per.problem selector. For example the segment for the first row of data will only be shown if 100 is selected for the bases.per.problem selector.
The showSelected.variable and showSelected.value mean to show the segment only if the value of showSelected.value is the current selection of the showSelected.variable selector. For example the segment for the first row of data will only be shown if 1 is selected for the size.100.problem.1peaks selector.

The new animint2 syntax uses parameters instead of aesthetics, so is much more concise:

geom_segment(
 showSelected=c(selector.name="peaks", "bases.per.problem"),
 clickSelects="problem.name")

Both showSelected and clickSelects should be character vectors. Named elements of the character vector are interpreted as the old variable/value aes, and un-named elements are interpreted as the old clickSelects/showSelected aes.

lazycipher/animint2