Access resources from CBCC website

• Create a new folder for storing your shiny apps.
• Place Hospitals.csv dataset in the new folder
• Place hospitalApp.zip in new folder and unzip
• Set your working directory to that folder
• Install shiny, refund, and refund.shiny libraries

# set working directory to be the folder you just created
setwd("/Users/juliawrobel/Documents/How to use Shiny/")

# install libraries
install.packages("shiny")
install.packages("refund.shiny")
install.packages("refund")

library(shiny)
runExample("01_hello")
runExample("03_reactivity")

Run the code above to see two examples of shiny apps

This app generates a simple interactive histogram.

# make sure your working directory is set to where the app is saved!
library(shiny)
runApp("myfirstshiny")

• Manually change color of the histogram

hist(x, breaks = bins, col = 'orange', border = 'magenta')

• Add radio buttons to let user pick a color

# on ui.R side 
radioButtons("fillColor", label = h3("Radio buttons"),
             choices = list("Red" = "firebrick", "Yellow" = "gold", 
                            "Blue" = "cornflowerblue"), 
             selected = "firebrick")

# on server.R side
hist(x, breaks = bins, col = input$fillColor, border = 'white')

• Make your own change! Try adding a text input widget for the title..

Data on 113 hospitals collected to determine whether infection surveillance and control programs have reduced the rates of nosocomial (hospital-acquired) infection.

hospital = read.csv("/Users/juliawrobel/Documents/
Presentations and Posters/CBCC/How to use Shiny/Hospital.csv")

hospital$MEDSCHL = factor(hospital$MEDSCHL, levels = 1:2, 
                          labels = c("yes", "no"))
hospital$REGION = factor(hospital$REGION, levels = 1:4, 
                         labels = c("NE", "NC", "S", "W"))

# variable designations
cats = c("MEDSCHL", "REGION")
conts = c("LOS", "AGE", "INFRISK", "CULT", "XRAY", 
          "BEDS", "CENSUS", "NURSE", "FACS")

We are going to build a shiny app for exploratory data analysis with the hospital data.

# server.R: change the following code to load Hospital.csv file
hospital = read.csv("/Users/juliawrobel/Documents
                    /Presentations and Posters/CBCC/
                    How to use Shiny/Hospital.csv")

# run the hospitals app
runApp("hospitalApp")

# ui.R change from this:
    ),
    tabPanel(title = "tab 2")

# to this:
    ),
    tabPanel(title = "tab 2"), # add comma here
    tabPanel(title = "name of tab 3") # add tabPanel function here

# ui.R
    # add widget
    selectInput("catVar", label = h4("Select a categorical variable"), 
                choices = as.list(cats), selected = 1)
    
    # add call to build plot
       h3("Bar plot of a categorical variable"),
       plotOutput("categoricalPlot")

# server.R
  output$categoricalPlot <- renderPlot({
    # get user selected categorical variable
    catVar <- hospital[, input$catVar] 
    
    # draw the barplot
    ggplot(hospital, aes(x = catVar)) + geom_bar()
  })

replace

# ui: change this
tabPanel(title = "tab 2"),

# to this
 tabPanel(title = "tab 2",
             sidebarLayout(
               sidebarPanel(
               ),
              
               mainPanel(
                 h3("Histogram of a continuous variable grouped 
                    by categorical variable"),
                 plotOutput("mixedPlot")
               )
             ) 
             ), # end of tab 2 

This time we use reactive expressions! Reactive expressions give you flexibility for re-using user selected inputs and make your code faster by selectively choosing what gets updated each time.

# add to server

contInput <- reactive({
  hospital[, input$contVar]
})
catInput <- reactive({
  hospital[, input$catVar]
})

  output$mixedPlot <- renderPlot({
    # draw the barplot
    ggplot(hospital, aes(x = contInput(), fill = catInput() )) +
      geom_histogram()
  })

Start with something simple. Some ideas/tips:

• plot of residuals from a linear model
• histogram that is grouped by a categorical variable
• scatter plot of two selected covariates
• refer to widgets page to learn how to code new widgets
• when debugging make sure your R code works before you try to put it in the shiny framework!

notice layouts, download buttons, multiple tabs, plotly

library(refund)
library(refund.shiny)

data(cd4)

# performs functional data analysis in refund package
SC = fpca.sc(cd4)

# visualizes the results of that analysis in refund.shiny
plot_shiny(SC)

• tutorials
• cool examples of what people have made with shiny
• widgets gallery
• layout guide
• build your shiny app as a function
• check out plotly for interactive plotting that intergrates with ggplot
• hosting and deploying shiny apps online
• reactive expression