Making a Covid-19 map in R using Shiny and Leaflet

Want to learn more about R and see it in action? Read on for a step-by-step tutorial – no previous coding experience necessary. This guest post was authored by Trent M. Ryan, PhD,  data scientist from the University of Melbourne’s Digital Studio.

Introduction

There have been a number of interactive Covid-19 maps released since the beginning of the pandemic. Johns Hopkins University was among the first of many institutions to create an interactive visualization illustrating the transmission dynamics of the coronavirus. In this tutorial, we will go combine a number of packages in R to create a similar interactive transmission map.

Required packages and data

No previous coding experience is necessary to follow along with this tutorial. However, you will need to install R and R Studio, which is a code editor for R to be able to run the code on your machine.

We will make use of a number of basic functions available in Shiny and Leaflet to create our map. Shiny is an R package that can create and launch web apps straight from the R Studio environment. Leaflet is an open-source JavaScript library used to create interactive maps. It is perhaps the most widely used and recognized package for mapping geospatial data.

The dataset for this tutorial comes from the World Health Organization. To keep things simple, the dataset only includes the number of cases over the course of the first 30 days of the pandemic. This dataset includes latitude and longitude information for each country with at least one positive case in the first thirty days of the pandemic.

Code

The first step is to install and load the libraries we need to create the map. We will use tidyverse to load and reshape the .csv dataset from wide to long format, leaflet and shiny to create and animate our map, and RColorBrewer to specify the color gradient we want to use on the map.

# install and load necessary libraries
# install.packages(c(tidyverse", "shiny", "leaflet", "RColorBrewer"))
library(tidyverse)
library(shiny)
library(leaflet)
library(RColorBrewer)

# load dataset
ncov <- read_csv("coronavirus_data.csv")

# turn to data.frame and reshape
ncov_d <- as.data.frame(ncov)
ncov_l <- reshape(ncov_d,
                  direction = "long",
                  varying = list(names(ncov_d)[6:35]),
                  v.names = "value",
                  idvar = c("country", "slat", "slon", "elat", "elon"),
                  timevar = "day")

# color palette
reds <- colorNumeric("Reds", domain = NULL)

Next, we need to specify the user interface properties (e.g., time slider and looping animation).

# specify shiny user interface and time slider
ui <- bootstrapPage(tags$style(type = "text/css", "html, body, .leaflet {width:100%; height:100%}"),
                    leafletOutput("map", width = "100%", height = "100%"),
                    # position and properties of the time slider
                    absolutePanel(bottom = 10, right = 300, draggable = TRUE,
                                  # slider title, step increments, and ticks
                                  sliderInput("integer", "Days since first reported case as of January 21, 2020:",ticks = FALSE, min = min(ncov_l$day), max = max(ncov_l$day), value = 1:30, step = 1,
                                              animate = animationOptions(interval = 1000, loop = TRUE))))

We will then need to specify the properties for the back end, including the background map and the reactive layer. For a complete explanation of how ui and server interact to create the map, refer to the shiny web documentation.

# shiny server input/output
server <- function(input, output, session) {
  filteredData <- reactive({
    ncov_l %>%
      filter(day >= input$integer[1] & day <= input$integer[2])
  })
  output$map <- renderLeaflet({
    reds <- colorNumeric("Reds", domain = NULL)
    leaflet(ncov_l) %>%
      addProviderTiles(providers$CartoDB.DarkMatterNoLabels) %>%
      # set boundaries for map
      fitBounds(lng1 = min(ncov_l$elon), 
                lat1 = min(ncov_l$elat), 
                lng2 = max(ncov_l$elon), 
                lat2 = max(ncov_l$elat)) %>%
      # add legend for the map
      addLegend("bottomleft", pal = reds, values = ~value,
                title = "Confirmed Cases <br> (data: WHO.int)",
                opacity = 0.5,
                bins = 4)
  })
  observe({
    leafletProxy("map", data = filteredData()) %>%
      clearMarkers() %>%
      addCircleMarkers(lng = ~elon,
                       lat = ~elat,
                       radius = ~log(value) * 3.5,
                       weight = 1,
                       opacity = 10,
                       color = ~ifelse(value > 0, reds(value), NA),
                       popup = ~paste0(country, ",  ", value, " cases"))
  }) 
}

The last step is to run the app.

shinyApp(ui, server, options = list(height = 550))      

Conclusion

The app is live! To see the day-to-day increase in transmissions, we can align the two slider markers and play the app. The app will loop once it reaches the day 30. The code above will work with any dataset that has latitude and longitude information and a time-varying frequency variable. And that’s it!

 

Learn more

The post was authored by Trent M, Ryan, PhD. Trent is a data scientist working with Social and Cultural Informatics Platform (SCIP) and the Digital Studio, Faculty of Arts, University of Melbourne. Explore more resources created by the SCIP team, including this recording of their recent webinar on how SCIP works with researchers at the University of Melbourne. 


Leave a Reply

Your email address will not be published. Required fields are marked *

Archives