Why?

• Deeper exploration of data than static plots (zoom, highlight relevant data points, hide irrelevant data)
• Immediate feedback on how data/figures/results change when inputs are modified
• User becomes an active participant in the analysis
• Exploratory analysis, teaching, reporting

A word of caution: if the interactivity doesn’t add anything to your visualization, don’t do it!

Recent explosion in tools for interactive visualizations in R:

• networkD3: Interactive network graphs with d3
• leaflet: add layers on top of maps
• googleVis: use Google Chart tools from R
• htmlwidgets: Javascript visualization libraries in R
• rCharts: R interface to multiple javascript charting libraries

• crosstalk: add-on for cross-widget interactions

• plotly: easily convert ggplot2 figures to interactive plots

• R library that allows you to turn standard R code into interactive web applications
• Two-way communication: R scripts can be rerun based on inputs from the app
• Requires server, or running locally in R

library(tidyverse)
p <- ggplot(iris, aes(x=Sepal.Length, y=Sepal.Width, 
                      color=Species, shape=Species)) + 
  geom_point(size=6, alpha=0.6)
p

library(plotly)
ggplotly(p, tooltip = c("y", "x"))

ggplotly(p, tooltip = c("y", "x", "colour"))

p <- ggplot(iris, aes(x=Sepal.Length, y=Sepal.Width, color=Species, 
                      shape=Species, label=Petal.Width)) + 
  geom_point(size=6, alpha=0.6)
ggplotly(p, tooltip = c("y", "x", "colour", "label"))

p2 <- ggplot(iris, aes(x=Species, y=Sepal.Length, 
                       fill=Species)) +  geom_boxplot() 
p2

ggplotly(p2)

library(d3heatmap)
d3heatmap(mtcars, scale="column", colors="Blues")

Example files:

• http://rpubs.com/andreamrau/310839 (open in Google Chrome)

• https://andreamrau.shinyapps.io/flexdashboard_example

• https://andreamrau.shinyapps.io/flexdashboard_example2

• Standalone HMTL files
• Host on website
• Embedded Rmarkdown files
• Post plotly graphics on plot.ly account (private or public URL)
• Embedded in Shiny applications
• F1000 research now allows for (plotly) interactive graphics! https://f1000research.com/articles/6-1396/v1#f6
• RPubs: Publish directly from RStudio

• Share the raw files for your Shiny app (.Rmd or ui.R + server.R files): requires user to have R/Shiny installed on their own computer and access to relevant data files

library(shiny)
runApp("...")
runURL("...")
runGitHub("...")

• Share Shiny app as a webpage: most user-friendly, local installation of R not required, app navigation through web browser

  • Web hosting yourself
  • Shinyapps.io: RStudio's hosting service with free → starter ($9/month) → basic ($39/month)→ standard ($99/month) → professional ($299/month) plans
    • Free plan limited to 5 apps, 25 active hours/month, no password protection
  • Shiny Server: companion program for Linux servers that builds a web server to host Shiny apps (free, open source)
  • Shiny Server Pro: paid version ($10k/year but 50% off discount for academic institutions) with password authentication, encryption, …

• Lots of recent development in the R community has made it very easy to incorporate interactivity into data analyses

• Out-of-box tools allow for easy initiation, but many specialized packages exist and lots of customization possible if some knowledge of Javascript, CSS, D3, …

• Sharing apps is still a bit of a sticking point: depends on intended audience, required computing power, number of expected users, …
  • What could/should INRA's strategy be? How many potential Shiny developers/users at INRA?
  • Invest in a (pro?) Shiny server to be installed on a platform server or an AWS instance with Shiny server?
  • Professional Shinyapps.io plan to be shared among users?

• http://rmarkdown.rstudio.com/flexdashboard/examples.html
• http://www.htmlwidgets.org
• https://plot.ly/r