Extending accessibility of open-source statistical software to the masses A shiny case study

Brandon LeBeau

2016-10-07

R

• R is an open source statistical programming language.
  • Pros:
    • Common statistical procedures are found in R
    • Can extend functionality with packages/functions
  • Cons:
    • Need to be comfortable with code

Flexibility of R

• R is powerful and flexible due to the many user written packages.
• However, to capture this flexibility:
  • users need to be comfortable with programming
  • users need to find the package
  • users need to understand package specific syntax

R package documentation and examples

https://www.rdocumentation.org/packages/dplyr/versions/0.5.0/topics/summarise

Blog posts

https://blog.rstudio.org/2014/01/17/introducing-dplyr/

Vignettes

https://cran.rstudio.com/web/packages/dplyr/vignettes/introduction.html

Weaknesses of these types of documentations

• They still rely on user understanding and reading R code.
• Not interactive, although the user can copy and paste code into an R session.
• This type of documentation will not capture the nontraditional useR.
• Shiny is the path to the nontraditional useR.

What is Shiny

• Shiny is an open-source framework for creating applications viewed in a web browser with R.
• Shiny Examples:
  • http://shiny.rstudio.com/gallery/movie-explorer.html
  • https://gallery.shinyapps.io/drinkr/
  • http://wordbank.stanford.edu/analyses?name=item_trajectories

Advantages of Shiny

• User needs no R knowledge
• App is viewed in the browser so able to use
  • Javascript
  • HTML
  • CSS
• Multiple hosting options
• Flexible Output

Disadvantages of Shiny

• Need a R developer to create the app.
  • More difficult as the code is somewhat different compared to traditional R code.
  • Shiny uses reactive programming.

Components of Shiny

1. User Interface (ui.r)
   • What the user sees and interacts with
2. R Analysis (server.r)
   • The R code running behind the scenes

User Interface

• Simple user interface example from RStudio
  • http://shiny.rstudio.com/gallery/telephones-by-region.html

shinyUI(
  fluidPage(    
    titlePanel("Telephones by region"),
    sidebarLayout(      
      sidebarPanel(
        selectInput("region", "Region:", 
                    choices = colnames(WorldPhones)),
        hr(),
        helpText("Data from AT&T (1961) The World's Telephones.")
      ),
      
      mainPanel(
        plotOutput("phonePlot")  
      )
    )
  )
)

Server File

• The server file for RStudio example
  • http://shiny.rstudio.com/gallery/telephones-by-region.html

shinyServer(function(input, output) {
  
  output$phonePlot <- renderPlot({
    
    barplot(WorldPhones[ , input$region] * 1000, 
            main = input$region,
            ylab = "Number of Telephones",
            xlab = "Year")
  })
})

Case Study

• pdfsearch
  • Note, you may need rtools to install this package.
• This following commands will run the pdfsearch shiny application locally.
  • Note, the following packages are required: shiny, shinydashboard, pdfsearch, DT https://github.com/lebebr01/pdfsearch

install.packages('devtools')
devtools::install_github('lebebr01/pdfsearch')
pdfsearch::run_shiny()

Case Study 2

• simglm
  • Note, need the following packages: shiny, shinydashboard, DT, simglm, ggplot2, lme4, highcharter https://github.com/lebebr01/simglm

devtools::install_github('lebebr01/simglm')
simglm::run_shiny()

Conclusions

• Shiny can give useRs an interactive framework to try out an R package.
• Benefits include
  • interactivity
  • no errors (for well developed Shiny applications)
  • no need to learn R or package specific syntax
  • only need a browser, no need to have R install locally when hosted on a server.

Questions?

• Twitter: @blebeau11
• Website: http://brandonlebeau.org
• Slides: http://brandonlebeau.org/2016/10/07/canam.html
• GitHub: http://github.com/lebebr01