Shiny Workshop

Shiny for R

• Introduction to Shiny
• Gapminder data
• Running example: visualising the gapminder data
• Deploying Shiny apps

Introduction to Shiny

• Shiny lets us easily and quickly build web apps using R
• Can use pretty much any output from R within our app
• And control it using widgets

  For example:

• Movie explorer
• Bus dashboard
• PK modelling
  

• Shiny gallery

Why (not?) use Shiny

• Fast to develop
• Use R’s libraries easily -including e.g. leaflet for mapping
• Many built in widgets
  • And packages for others

• Awkward to scale
• Non-free pro version needed for some server features

Gapminder

Workshop Aims

• To use Shiny to make an interactive plot
  • Select / animate the year
  • Only plot data for selected continents
  • Deploy the app to shinyapps.io
  • (Possible extensions included in “going further”)

Getting started

• Run rstudio:

rstudio

• gapminder.rds contains the gapminder data

• workshopFunctions.R contains the functions we will use.
  • These are mostly to make it easier to make the plots, and to avoid this turning into a ggplot workshop.

• codeExample.R shows how to use the plotting functions.

• The course website https://uomresearchit.github.io/RSE18-shiny-workshop/ contains all the exercises, links to solutions and more detail than these slides.

Interactive exercises

• The mawdsley folder is a git repository
• Solutions are stored as tagged commits for code in the worked_example directory
  • e.g git:01_defaultapp
  • Link takes you to github’s commit page
  • Checkout versions with git checkout 01_defaultapp
• Make your own Shiny app in another directory

Getting started

• Make a new Shiny app and check it works: git:01_defaultapp
  • File, New File, Shiny Web App
  • Name: gapminder
  • Single file
• Run App using toolbar or Ctrl+Shift+Enter

Anatomy of a Shiny App - user interface

ui <- fluidPage(
    
    # Application title
    titlePanel("Old Faithful Geyser Data"),
    
    # Sidebar with a slider input for number of bins 
    sidebarLayout(
        sidebarPanel(
            sliderInput("bins", ...
        ),
        
        
        mainPanel(
           plotOutput("distPlot")
        )
    )
)

input$bins is a source

output$distPlot is an endpoint

Anatomy of a Shiny App - server

server <- function(input, output) {
    
    output$distPlot <- renderPlot({
        # generate bins based on input$bins from ui.R
        x    <- faithful[, 2] 
        bins <- seq(min(x), max(x), length.out = input$bins + 1)
        
        # draw the histogram with the specified number of bins
        hist(x, breaks = bins, col = 'darkgray', border = 'white')
    })
}

Shiny handles the dependencies between sources and endpoints, and only updates what is needed - Reactive programming.

Exercise - getting ready for the rest of the course.

• Copy the gapminder data (gapminder.rds) and workshopFunctions.R to your app directory

• Modify the app to load the ggplot2 and dplyr libraries, gapminder data, and workshopFunctions.R

• Check your app still works

git:02_loaddata

(Course notes: Getting started)

More widgets

• Many common widget types are built into Shiny see: https://shiny.rstudio.com/gallery/widget-gallery.html
• Can be extended using additional packages (e.g. https://github.com/dreamRs/shinyWidgets)
• or by writing your own.

Exercise

• Add a checkboxGroupInput() widget to select the continents we wish to display on the graph
• Change the title of the app in titlePanel() to something sensible

(This won’t do anything yet - we’ll connect it to the graph shortly)

git:03_continentwidget

(Course notes: User interface design)

Exercises - Putting it all together

• Replace the histogram with a gapminder plot, using produceGapminderPlot() git:04_gapminderplot
• See codeExample.R
• Replace the bins widget with a widget to select the year git:05_yearwidget
• Use the outputs of the widgets to make the graph interactive git:06_interactive

(Course notes: Putting it all together)

Deploying Shiny apps

• Run from R(Studio)

• https://shinyapps.io
  • Easy deployment from RStudio
  • Various pricing tiers (inc. free)
• Shiny server
  • Free and commercial versions
  • Authentication requires commercial version

(Course notes: Deploying Shiny apps)

Summary

• Define the UI and server functions
• Reactive programming handles the depencencies between elements
• Deploy apps using shinyapps.io or to a Shiny Server

Extra Slides

App layout

• The default apps sidebarLayout() is a good starting point.
• The fluidPage() layout gives us more control.

ui <- fluidPage(
  titlePanel("Gapminder visualisation"), # Will cover whole width
  plotOutput("gapminderPlot", click = "plotClick"), # Ditto
  fluidRow( 
    column(6, # column width
           sliderInput(inputId = "year", ....) # Things to include in the column
    ),
    column(6, # Next column
           checkboxGroupInput("continent", ....)# Things to include in second column
    )
  )
)

• The widths of each row of columns should sum to 12 git:15_layout