More than two map levels • leafdown

In this article, we demonstrate how to set up a leafdown map with three map levels and show that it is an natural extension of the two map level case. (More than two map levels is supported from leafdown version 1.1.)

Let’s first load the libraries we are going to use for our app.

library(leafdown)
library(leaflet)
library(shiny)
library(dplyr)
library(shinycssloaders)
library(shinyjs)
library(dplyr)
library(raster)

(Note that the shinyjs package is loaded for some automatic warning messages that the leafdown map can return to the user of the shiny app.)

SpatialPolygonsDataFrames

As usual, we first create a list of SpatialPolygonsDataFrames (spdfs) for the regions we want to display on our map. (Further details can be found in the Introduction article.)

list (spdfs_list)
│   
└───spdf (spdf of first map level)
│
└───spdf (spdf of second map level)
│
└───spdf (spdf of third map level)

# Germany
ger0 <- raster::getData(country = "Germany", level = 0)
ger1 <- raster::getData(country = "Germany", level = 1)
ger2 <- raster::getData(country = "Germany", level = 2)
# Austria
a0 <- raster::getData(country = "Austria", level = 0)
a1 <- raster::getData(country = "Austria", level = 1)
a2 <- raster::getData(country = "Austria", level = 2)
# Switzerland
ch0 <- raster::getData(country = "Switzerland", level = 0)
ch1 <- raster::getData(country = "Switzerland", level = 1)
ch2 <- raster::getData(country = "Switzerland", level = 2)

spdfs_list <- list(rbind(aut0, ch0, ger0), rbind(aut1, ch1, ger1), rbind(aut2, ch2, ger2))

Data

For this app, we simply generate random target values y.

set.seed(20220106)
# Simulate data
data_sim_y_level_3 <- spdfs_list[[3]]@data
data_sim_y_level_3$y <- round(rnorm(nrow(data_sim_y_level_3), 1e2, sd = 5e2), 0)
data_sim_y_level_2 <- data_sim_y_level_3 %>% group_by(NAME_0, NAME_1) %>% summarise(y = sum(y))
data_sim_y_level_1 <- data_sim_y_level_2 %>% group_by(NAME_0) %>% summarise(y = sum(y))

# Assign map levels
data_sim_y_level_3$level <- 3
data_sim_y_level_2$level <- 2
data_sim_y_level_1$level <- 1

# Assign area names
data_sim_y_level_3$area <- data_sim_y_level_3$NAME_2
data_sim_y_level_2$area <- data_sim_y_level_2$NAME_1
data_sim_y_level_1$area <- data_sim_y_level_1$NAME_0

# Combine data of map levels
data_sim_y <- rbind(
  data_sim_y_level_3[, c("area", "y", "level")],
  data_sim_y_level_2[, c("area", "y", "level")],
  data_sim_y_level_1[, c("area", "y", "level")]
)

head(data_sim_y)
#>                  area    y level
#> 1          Eisenstadt 1086     3
#> 2 Eisenstadt Umgebung  404     3
#> 3             Güssing  502     3
#> 4         Jennersdorf  752     3
#> 5         Mattersburg  329     3
#> 6     Neusiedl am See  184     3

Joining map levels

In order for a leafdown map to know which shapes of a lower map level belong to the shapes in the upper map level, we need to specify how the data of the spdfs of different map levels need to be joined. This is defined by the join_map_levels_by argument, which is a named vector of length length(spdfs_list) - 1, whereby

the first (named) element specifies how to join the first and the second map level
the second (named) element specifies how to join the second and the third map level
and so on.

The name of an element specifies the join column in the respective upper map level and the actual element the join column of the lower map level.

Let’s have a look at the data of our the spdfs in spdf_list:

head(spdfs_list[[1]]@data)
#>   GID_0      NAME_0
#> 1   AUT     Austria
#> 2   CHE Switzerland
#> 3   DEU     Germany

head(spdfs_list[[2]]@data[, c("GID_0", "NAME_0", "GID_1", "NAME_1")])
#>   GID_0  NAME_0   GID_1           NAME_1
#> 1   AUT Austria AUT.1_1       Burgenland
#> 2   AUT Austria AUT.2_1          Kärnten
#> 3   AUT Austria AUT.3_1 Niederösterreich
#> 4   AUT Austria AUT.4_1   Oberösterreich
#> 5   AUT Austria AUT.5_1         Salzburg
#> 6   AUT Austria AUT.6_1       Steiermark

In this example, map levels 1 and 2 can be joined by their column “GID_0”. So the name of first element is “GID_0” (the respective column name of the first map level) and the first element is also “GID_0” (the respective column name of the first map level):

head(spdfs_list[[3]]@data[, c("GID_0", "NAME_0", "GID_1", "NAME_1", "GID_2", "NAME_2")])
#>   GID_0  NAME_0   GID_1     NAME_1     GID_2              NAME_2
#> 1   AUT Austria AUT.1_1 Burgenland AUT.1.2_1          Eisenstadt
#> 2   AUT Austria AUT.1_1 Burgenland AUT.1.1_1 Eisenstadt Umgebung
#> 3   AUT Austria AUT.1_1 Burgenland AUT.1.3_1             Güssing
#> 4   AUT Austria AUT.1_1 Burgenland AUT.1.4_1         Jennersdorf
#> 5   AUT Austria AUT.1_1 Burgenland AUT.1.5_1         Mattersburg
#> 6   AUT Austria AUT.1_1 Burgenland AUT.1.6_1     Neusiedl am See

We see, that map levels 2 and 3 can be joined by their column “GID_1”.

This translates into the following join_map_levels_by argument:

my_leafdown <- Leafdown$new(
  spdfs_list, "leafdown", input, join_map_levels_by = c("GID_0" = "GID_0", "GID_1" = "GID_1")
)

Shiny App

Preparation

library(leafdown)
library(leaflet)
library(shiny)
library(dplyr)
library(shinycssloaders)
library(shinyjs)
library(dplyr)
library(raster)

# Germany
ger0 <- raster::getData(country = "Germany", level = 0)
ger1 <- raster::getData(country = "Germany", level = 1)
ger2 <- raster::getData(country = "Germany", level = 2)
# Austria
a0 <- raster::getData(country = "Austria", level = 0)
a1 <- raster::getData(country = "Austria", level = 1)
a2 <- raster::getData(country = "Austria", level = 2)
# Switzerland
ch0 <- raster::getData(country = "Switzerland", level = 0)
ch1 <- raster::getData(country = "Switzerland", level = 1)
ch2 <- raster::getData(country = "Switzerland", level = 2)

# load the shapes for the three levels
spdfs_list <- list(rbind(aut0, ch0, ger0), rbind(aut1, ch1, ger1), rbind(aut2, ch2, ger2))

# Simulate some data
set.seed(20220106)
data_sim_y_level_3 <- spdfs_list[[3]]@data
data_sim_y_level_3$y <- rnorm(nrow(data_sim_y_level_3), 1e2, sd = 5e2)
data_sim_y_level_2 <- data_sim_y_level_3 %>% group_by(NAME_0, NAME_1) %>% summarise(y = sum(y))
data_sim_y_level_1 <- data_sim_y_level_2 %>% group_by(NAME_0) %>% summarise(y = sum(y))

data_sim_y_level_3$level <- 3
data_sim_y_level_2$level <- 2
data_sim_y_level_1$level <- 1

data_sim_y_level_3$area <- data_sim_y_level_3$NAME_2
data_sim_y_level_2$area <- data_sim_y_level_2$NAME_1
data_sim_y_level_1$area <- data_sim_y_level_1$NAME_0

data_sim_y <- rbind(
  data_sim_y_level_3[, c("area", "y", "level")],
  data_sim_y_level_2[, c("area", "y", "level")],
  data_sim_y_level_1[, c("area", "y", "level")]
)
data_sim_y$y <- round(data_sim_y$y, 0)

UI

ui <- fluidPage(
  mainPanel(
    # set the background of the map-container to be white
    tags$head(
      tags$style(HTML(".leaflet-container { background: #fff; height: 100%}")),
      # workaround for the NA in leaflet legend see https://github.com/rstudio/leaflet/issues/615
      tags$style(HTML(".leaflet-control div:last-child {clear: both;}"))
    ),
    # we need shinyjs for the leafdown map
    useShinyjs(),
    fluidRow(
      # the two buttons used for drilling
      actionButton("drill_down", "Drill Down"),
      actionButton("drill_up", "Drill Up"),
      # the actual map element
      withSpinner(leafletOutput("leafdown", height = 800), type = 8)
    )
  )
)

Server

# Little helper function for hover labels
create_labels <- function(data, map_level) {
  labels <- sprintf(
    "<strong>%s</strong><br/>%g</sup>",
    data[, paste0("NAME_", map_level - 1)], data$y
  )
  labels %>% lapply(htmltools::HTML)
}

server <- function(input, output) {

  # create leafdown object
  my_leafdown <- Leafdown$new(
    spdfs_list, "leafdown", input, join_map_levels_by = c("GID_0" = "GID_0", "GID_1" = "GID_1")
  )

  rv <- reactiveValues()
  rv$update_leafdown <- 0

  # observers for the drilling buttons
  observeEvent(input$drill_down, {
    my_leafdown$drill_down()
    rv$update_leafdown <- rv$update_leafdown + 1
  })

  observeEvent(input$drill_up, {
    my_leafdown$drill_up()
    rv$update_leafdown <- rv$update_leafdown + 1
  })

  data <- reactive({
    req(rv$update_leafdown)
    meta_data <- my_leafdown$curr_data
    curr_map_level <- my_leafdown$curr_map_level
    data_curr_map_level <- data_sim_y[data_sim_y$level == curr_map_level, ]
    join_col_lhs <- paste0("NAME_", curr_map_level - 1)
    data <- meta_data %>% left_join(data_curr_map_level, by = setNames("area", join_col_lhs))

    # add the data back to the leafdown object
    my_leafdown$add_data(data)
    data
  })

  # this is where the leafdown magic happens
  output$leafdown <- renderLeaflet({
    req(spdfs_list)
    req(data)

    data <- data()
    labels <- create_labels(data, my_leafdown$curr_map_level)
    # draw the leafdown object
    my_leafdown$draw_leafdown(
      fillColor = ~leaflet::colorNumeric("Greens", data$y)(data$y),
      weight = 3, fillOpacity = 1, color = "grey", label = labels
    )
  })
}

Run App

shinyApp(ui, server)