Interactive data visualization: shiny and ggvis

• A web application framework for R

  • can create data driven web application
  • knowledge of HTML, JavaScript and CSS are not necessary

• Can be used to create data product

• D3 plots can be included in shiny application

• Can be easily shared in various ways

  • deploy on RStudio server
  • deploy on your own website

install.packages("shiny")
library(shiny)

• Click here to view some example

• You need two R source files saved in a folder say myShinyApp. Source files should be named

  • server.R
  • ui.R

• server.R contains the R codes to

  • generate the plots
  • present tables of results

• ui.R contains the R codes to

  • design the user interface for facilitating the user interactivity with the plots
  • to add typical widgets such as slider, check box, select box etc.

• Two ways to run your application

  • from command prompt type runApp("path-to/myShinyApp")
  • from RStudio, click on Run App

library(shiny)

# Define server logic required to draw a histogram
shinyServer(function(input, output) {

  # Expression that generates a histogram. The expression is
  # wrapped in a call to renderPlot to indicate that:
  #
  #  1) It is "reactive" and therefore should re-execute automatically
  #     when inputs change
  #  2) Its output type is a plot

  output$distPlot <- renderPlot({
    x    <- faithful[, 2]  # Old Faithful Geyser data
    bins <- seq(min(x), max(x), length.out = input$bins + 1)

    # draw the histogram with the specified number of bins
    hist(x, breaks = bins, col = 'darkgray', border = 'white')
  })
})

# Define UI for application that draws a histogram
shinyUI(fluidPage(

  # Application title
  titlePanel("Hello Shiny!"),

  # Sidebar with a slider input for the number of bins
  sidebarLayout(
    sidebarPanel(
      sliderInput("bins",
                  "Number of bins:",
                  min = 1,
                  max = 50,
                  value = 30)
    ),

    # Show a plot of the generated distribution
    mainPanel(
      plotOutput("distPlot")
    )
  )
))

• Click here to view the shiny widget gallary

• The default shiny theme is bootstrap which is usually loaded from downloaded bootstrap.css

shinyUI(fluidPage(theme = "bootstrap.css",

  titlePanel("My Application")

  # application UI              
))

• Some themes can be viewed from
  • http://bootswatch.com/
  • source codes can be viewed and copied depending on which theme you prefer

• Let us create an application using shiny

• We will create an application to explore US crime data from 1969 to 2005

• The application that we made is already hosted on Github. To launch this application, run the following codes from R command prompt

library(shiny)
shiny::runGitHub('usa-crime', 'mamajumder')

• Before running this make sure packages ggplot2, dplyr, scales and ggmap are already installed.

• The source codes and the data for this application can be downloaded from
  https://github.com/mamajumder/usa-crime

• The application is deployed on RStudio. To view the application we created visit
  https://mahbub.shinyapps.io/usa-crime/

• Some interactive plots can be generated using package ggvis
  • still being developed
  • can be used in shiny application
  • similar to ggplot2 but designed for dynamic web graphics
  • uses chain operation %>% for multiple layers

install.packages("ggvis")
library(ggvis)

• User inputs are passed through some input functions

• Input functions are
  • input_slider(): slider input
  • input_checkbox(): a check-box
  • input_checkboxgroup(): a group of check boxes
  • input_numeric(): a spin box
  • input_radiobuttons(): pick one from a set options
  • input_select(): create a drop-down text box
  • input_text(): arbitrary text input
• Example

mtcars %>% ggvis(x= ~wt) %>%
  layer_densities(
    stroke := input_radiobuttons(c("Purple","Orange","steelblue"), 
                                 label="Line color"),
    fill := input_select(c("Purple","Orange","steelblue"), 
                         label="Fill color")
    )

library(ggvis)
mtcars %>% ggvis(x = ~wt) %>%
    layer_densities(
      adjust = input_slider(.1, 2, value = 1, step = .1, label = "Bandwidth adjustment"),
      kernel = input_select(
        c("Gaussian" = "gaussian",
          "Epanechnikov" = "epanechnikov",
          "Rectangular" = "rectangular",
          "Triangular" = "triangular",
          "Biweight" = "biweight",
          "Cosine" = "cosine",
          "Optcosine" = "optcosine"),
        label = "Kernel")
    )

• ggvis can do limited things. But what it does, it requires less effort.
  • For more control use shiny

• To view the shiny application we created to explore US crime data visit
  https://mahbub.shinyapps.io/usa-crime/

• Shiny web site
  http://shiny.rstudio.com/

• Shiny cheat sheet, a very useful reference guide. Check for updates.
  http://shiny.rstudio.com/articles/cheatsheet.html

• Overview of R package ggvis
  http://ggvis.rstudio.com/