Shiny

Package installation

install.packages('shiny', 'portalr')

Starting a new App

• New -> Shiny Web App
• Name of App
• Location
• Creates a small example app

Running the App

• Click the Run App button
• This example app makes histograms of times between eruptions of the Old Faithful Geyser
• By default runs in a separate window, but can change to viewer or browser

Basic structure

• UI - User Interface
  • Describes what the user sees and how they interact with it
• Server
  • Performs calculations, makes graphs, etc.
• Call to shinyApp actually create the Shiny App

Our project

• Make an app that displays time-series of mammal community dynamics at Portal
• Let the use choose which species then want to visualize

UI

• titlePanel controls the title of the page
• sidebarLayout controls what is shown in the sidebar

• mainPanel controls what is shown in the main page

• Change the title of the page to “Portal Rodent Species Dynamics”

titlePanel("Portal Rodent Species Dynamics")

• Run App

• Change the sidebar to let us select from the available species

Loading data

• First get the data using portalr

library(portalr)
library(dplyr)
library(ggplot2)

abundances <- abundance(shape = "long", time = "date", clean = FALSE)

• Then get a list of the species

species_list <- unique(abundances$species)

• Doesn’t need to update during session so put it outside of server and ui

Changing Sidebar

• Now change the sidebarPanel widget to a selectInput box
  • Drop down to select one or more options
  • Arguments
    • Input ID - variable name for storing the selected value
    • label - the header for this piece of the UI
    • choices - the values the user can choose from

selectInput("species",
            "Species",
            species_list)

• Run App

Server

• The server function is where calculations are done and graphs are made

Changing the graph

• renderPlot function lets us create a plot that can be included in the UI
• First filter the datset using dplyr, then plot with ggplot
• input$species indicates the value collected from the selectInput drop-down

output$distPlot <- renderPlot({
    filtered_abundances <- abundances %>% 
      filter(species == input$species)
    ggplot(filtered_abundances, aes(x = censusdate, y = abundance)) +
      geom_line()

• This plot will be made in the main panel because plotOutput("distPlot") displays the output of `output$distPlot

• Run App

Selecting multiple species

• Can also choose multiple species at once
• Change the UI to allow selecting more than one species
• Change the server to filter and graph multiple species

selectInput("species",
            "Species",
            species_list,
            multiple = TRUE)

filter(species %in% input$species)

ggplot(filtered_abundances, aes(x = censusdate, y = abundance, color = species))

• Run App

Change the date range

• Allow focus on narrower date range
• Update data code to get min and max dates

min_date <- min(abundances$censusdate)
max_date <- max(abundances$censusdate)

• Update the UI to have a slider to specify the date range

sliderInput("dates",
            "Date Range",
            min = min_date,
            max = max_date,
            value = c(min_date, max_date))

• Update the server to filter based on the dates in this input

filter(censusdate >= input$dates[1], censusdate <= input$dates[2])

Sharing your app

• Anyone can run your file in RStudio
• Can be run directly from GitHub

shiny::runGitHub('portal-explorer', 'weecology')

• Can also publish the app on the web
• Need an account on https://shinyapps.io
• Setup and the click Publish Application button

https://ethanwhite.shinyapps.io/portal-explorer/

Add an optional smoother (optional extension for longer time-periods)

• Allow us to look at trends in the time-series

• Add a checkbox to the UI

checkboxInput("smoother", 
            "Smoother")

• Add an optional smoother to out plot
• Make the basic plot then use a conditional to add (or not) the smoother

ts_plot <- ggplot(filtered_abundances, aes(x = censusdate, y = abundance, color = species)) +
    geom_line()
if (input$smoother){
    ts_plot <- ts_plot + geom_smooth()