Shiny App

Hosting Data Apps

With the UI and the server function, we can run the Shiny app locally:

shinyApp(ui = ui, server = server)

Single file app

We can combine the code so far into a single file, e.g. called app.R:

library(shiny)
ui <- fluidPage(
mainPanel(
sliderInput("obs",
"Number of observations",
min = 1,
max = 5000,
value = 100),
plotOutput("distPlot")
)
)
server <- function(input, output) {
output$distPlot <- renderPlot({
dist <- rnorm(input$obs)
hist(dist,
col="purple",
xlab="Random values")
})
}
shinyApp(ui = ui, server = server)

What's new here is that we passed ui and server to shinyApp() which returns the Shiny app object. This is passed to runApp() via the implicit print method to run the app.

The runApp() function can also take the app.R file as argument because the file returns the app object: runApp("app.R").

Multiple file app

An alternative to having everything for our app in a single app.R file, we can specify a directory where the app.R file can be found. However, when a directory is specified, we can have the Shiny app components in separate files. This usually helps in maintaining the app when it grows more complex.

It is good practice to create a global.R file with all the prerequisites required by the app, i.e. loading libraries, scripts, data sets, running data processing or defining functions. In our simple case we only load the shiny package. Thus in our global.R file we'll have the following line:

library(shiny)

The ui.R file contains the UI definition:

ui <- fluidPage(
mainPanel(
sliderInput("obs",
"Number of observations",
min = 1,
max = 5000,
value = 100),
plotOutput("distPlot")
)
)

Similarly, we have a server.R file defining the server function:

server <- function(input, output) {
output$distPlot <- renderPlot({
dist <- rnorm(input$obs)
hist(dist,
col="purple",
xlab="Random values")
})
}

We can now point runApp() to this directory.

Deployment

Next we will look at Shiny app deployment. Besides the officially recommended deployment options, such as shinyapps.io or the open source Shiny server, there are other possibilities. One particular option that we will review in depth is the deployment of Shiny apps via Docker and ShinyProxy. Read on!

Hosting Data Apps

Last updated on by Peter Solymos