• Importance of visualising data
• Interactive graphics in R
  • googleVis
  • htmlwidgets
  • ggvis
• Shiny
• Your turn

It may not be long before presentations with static plots are as old fashioned as whipping out your overhead transparencies.

install.packages("googleVis")
require(googleVis)
demo(WorldBank, package = "googleVis")

• Markus Gesmann created the googleVis package which accesses the Google Chart API directly from R

• The functions of the package allow users to visualise data with the Google Chart Tools without uploading their data to Google

• The output of googleVis functions is html code that contains the data and references to JavaScript functions hosted by Google

Using the Google Visualisation API with R, The R Journal, 3(2):40-44, December 2011 and googleVis package vignette

developers.google.com/chart

Get it on GitHub

install.packages("devtools")
devtools::install_github("garthtarr/mplot")
require(mplot)

... or get it on CRAN

install.packages("mplot")

• vis() for variable inclusion and model stability plots
• af() for the adaptive fence
• bglmnet() for bootstrapping glmnet
• mplot() for an interactive Shiny interface

Tarr G, Mueller S and Welsh AH (2015). “mplot: An R package for graphical model stability and variable selection.” arXiv:1509.07583 [stat.ME].

require(mplot)
lm.d = lm(y ~ ., data = diabetes)
vis.d = vis(lm.d, B = 200)
plot(vis.d, which = "vip")

plot(vis.d, which = "boot", highlight = "ltg")

af.d = af(lm.d, B = 200, n.c = 100, c.max = 100)
plot(af.d)

• Requires internet access to work
• some charts require Flash
• not very customisable

htmlwidgets is a package that lets you relatively easily link R with JavaScript.

Some examples:

• networkD3: force directed networks, Sankey diagrams and tree networks
• edgebundleR: hierarchical edge bundle plots
• pairsD3: interactive scatter plot matrices
• dygraphs: charting time-series data
• leaflet: dynamic maps
• DT: interactive HTML data tables
• rpivotTable: build pivot tables to dynamically slice, dice and plot your data

hafen.github.io/htmlwidgetsgallery

• most will work without internet access
• customisable (sort of)
• active community
  • some packages are backed by RStudio
  • Kent Russel ( @timelyportfolio) has made one widget each week this year
• if you're keen you can make your own... it's relatively easy to adapt any of the multitude of D3 examples

install.packages("magrittr")
require(magrittr)
awesome_data =
  raw_interesting_data %>%
  transform(somehow) %>%
  filter(the_good_parts) %>%
  finalize

require(magrittr)
x = rnorm(10)
x %>% max

## [1] 1.405986

max(x)

## [1] 1.405986

x %>% sort

##  [1] -0.97010061 -0.74849748 -0.60678533 -0.38610363 -0.03629047
##  [6]  0.01187292  0.42919359  0.65649122  0.78735912  1.40598620

sort(x)

##  [1] -0.97010061 -0.74849748 -0.60678533 -0.38610363 -0.03629047
##  [6]  0.01187292  0.42919359  0.65649122  0.78735912  1.40598620

x %>% sort %>% round(1)

##  [1] -1.0 -0.7 -0.6 -0.4  0.0  0.0  0.4  0.7  0.8  1.4

round(sort(x), 1)

##  [1] -1.0 -0.7 -0.6 -0.4  0.0  0.0  0.4  0.7  0.8  1.4

x %>% sort %>% round(1) %>% plot

plot(round(sort(x), 1))

Similar to JavaScript's chaining:

d3.select("body")
    .append("p")
    .text("New paragraph!");

Compund assignment %<>%

x %<>% sort %>% round(1)
x

##  [1] -1.0 -0.7 -0.6 -0.4  0.0  0.0  0.4  0.7  0.8  1.4

Read more

• Web based incarnation of ggplot2 (svg or canvas)
• Enables simple interactivity (e.g. tooltips)
• Advanced interactivity requires an R instance in the background (e.g. using sliders to change parameters)

• Not as simple as changing ggplot() to ggvis()
• Under development - not quite production ready yet
• It can't facet (yet)
• Recently on twitter @hadleywickham commented "goal is for 2016 to be the year of ggvis"

Read more: ggvis.rstudio.com

devtools::install_github("ramnathv/rCharts")

• rcharts.io
• Documentation is (currently) extremely limited

plotly: Interactive Graphics via Plotly's JS Graphing Library

install.packages("plotly")

• plot.ly/r
• takes a ggplot 2 object and converts it to an interactive chart
• requires authentication

rggobi: Interface between R and GGobi

To get started look at Wickham, Lawrence, Lang, Swayne (2008). "An Introduction to rggobi", R_News, 8(2) 3-7.

install.packages("qtlcharts")

• kbroman.org/qtlcharts

I wanted a way to interactively explore the results of a mega-simulation study from my PhD thesis.

Try loading it from GitHub :

install.packages("huge")
shiny::runGitHub("robnetwork", "garthtarr")

Otherwise it's also hosted here: Shinyapps.io

• No special HTML or JavaScript knowledge required
• Share the power of R with non R users
• Empower your collaborators to answer their own questions

Shinyapps.io

There's a button in RStudio that lets you push it to their server.

GitHub

Host your app on GitHub and then use code like this:

shiny::runGitHub('robnetwork', 'garthtarr')

Host your own RStudio/Shiny server (more complicated)

More info

• Interactivity with base and ggplot2 graphics (e.g. clicking and brushing)

More info

• 2.5 hour long video
• 7 very comprehensive lessons
• shiny.rstudio.com/tutorial

1. Run through the lecture code then play with a Shiny app
2. Build a Shiny app